> ## Documentation Index
> Fetch the complete documentation index at: https://docs.deeda.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Google Gemini

> 466 usable capability rows, 394 documented parameters.

What you can do with **Google Gemini** through the harness, by feature
family: the capability rows you can rely on (linked to vendor docs) and
the **parameters** you can set for each. Configure parameters through the
workflow surfaces described in the [Workflow Schema](/harness/workflow-schema)
— model/turn/budget fields on `agent`, provider knobs under
`harness_config.sdk_settings.gemini`, and tool/sandbox policy at top level.

## How to read this page

Decision rules per capability state — what YOU (the reader/agent) should do:

| State badge            | Meaning                                                                                                                                                                                                                                                                             | Your action                                                                                                                                                                         |
| ---------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| *(none)* = implemented | Wired end-to-end through deeda primitives and live-probed.                                                                                                                                                                                                                          | Use freely.                                                                                                                                                                         |
| *model-dependent*      | Works only on a documented subset of the provider's models.                                                                                                                                                                                                                         | Pin a supporting `model` in your workflow; verify against the vendor docs link.                                                                                                     |
| *partial-compatible*   | One half of the contract is proven, the other is not (yet). Patterns: request shape accepted but execution unproven at scale; wired in deeda but live probe pending; vendor does it but no deeda knob exists; works on one dispatch path (e.g. CLI) but not another (e.g. raw API). | Usable — but treat the **Parameters table as the entire contract**. Do not assume vendor-parity beyond what is listed. If your task depends on the unproven half, test small first. |

Parameter table columns: **Parameter** is the provider-side path the harness
maps for you (you set it via workflow fields or `harness_config.sdk_settings.<provider>`
knobs — see [Runtimes](/harness/runtimes) for which knobs exist). **Default**
applies when you say nothing. **Allowed** is exhaustive — values outside it
fail validation. **Risk** is deeda's policy weight: `high` parameters
generally require elevated `tool_policy`/`sandbox_profile` and may trigger
human approval on hot-reload.

## Agent Development Kit

**Capabilities** (46/55 rows usable):

* `gemini.adk.a2a.protocol` (agent.adk\_multi) — [vendor docs](https://google.github.io/adk-docs/agents/multi-agents/)
* `gemini.adk.agent.custom` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/agents/custom-agents/)
* `gemini.adk.agent.llm` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/agents/llm-agents/)
* `gemini.adk.agent.loop` (agent.adk\_workflow) — [vendor docs](https://google.github.io/adk-docs/agents/workflow-agents/loop-agents/)
* `gemini.adk.agent.parallel` (agent.adk\_workflow) — [vendor docs](https://google.github.io/adk-docs/agents/workflow-agents/parallel-agents/)
* `gemini.adk.agent.sequential` (agent.adk\_workflow) — [vendor docs](https://google.github.io/adk-docs/agents/workflow-agents/sequential-agents/)
* `gemini.adk.artifact.in_memory` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/artifacts/)
* `gemini.adk.callback.after_agent` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/callbacks/)
* `gemini.adk.callback.after_model` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/callbacks/)
* `gemini.adk.callback.after_tool` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/callbacks/)
* `gemini.adk.callback.before_agent` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/callbacks/)
* `gemini.adk.callback.before_model` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/callbacks/)
* `gemini.adk.callback.before_tool` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/callbacks/)
* `gemini.adk.callbacks` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/callbacks/)
* `gemini.adk.cli.api_server` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/runtime/)
* `gemini.adk.cli.eval` (eval.adk) — [vendor docs](https://google.github.io/adk-docs/evaluate/)
* `gemini.adk.cli.run` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/get-started/quickstart/)
* `gemini.adk.cli.web` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/get-started/quickstart/)
* `gemini.adk.eval.criteria` (eval.adk) — [vendor docs](https://google.github.io/adk-docs/evaluate/)
* `gemini.adk.evaluate` (eval.adk) — [vendor docs](https://google.github.io/adk-docs/evaluate/)
* `gemini.adk.events` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/events/)
* `gemini.adk.input_schema` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/agents/llm-agents/)
* `gemini.adk.loop_agent.max_iterations` (agent.adk\_workflow) — [vendor docs](https://google.github.io/adk-docs/agents/workflow-agents/loop-agents/)
* `gemini.adk.memory.in_memory` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/sessions/memory/)
* `gemini.adk.multi_agent` (agent.adk\_multi) — [vendor docs](https://google.github.io/adk-docs/agents/multi-agents/)
* `gemini.adk.output_key` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/agents/llm-agents/)
* `gemini.adk.output_schema` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/agents/llm-agents/)
* `gemini.adk.pattern.escalate` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/agents/llm-agents/)
* `gemini.adk.pattern.sub_agents` (agent.adk\_multi) — [vendor docs](https://google.github.io/adk-docs/agents/multi-agents/)
* `gemini.adk.pattern.transfer` (agent.adk\_multi) — [vendor docs](https://google.github.io/adk-docs/agents/multi-agents/)
* `gemini.adk.run_config.streaming` (agent.adk\_streaming) — [vendor docs](https://google.github.io/adk-docs/runtime/)
* `gemini.adk.runner` (agent.adk\_runtime) — [vendor docs](https://google.github.io/adk-docs/runtime/)
* `gemini.adk.session.database` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/sessions/session/)
* `gemini.adk.session.in_memory` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/sessions/session/)
* `gemini.adk.state.app_scope` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/sessions/state/)
* `gemini.adk.state.shared` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/sessions/state/)
* `gemini.adk.state.temp_scope` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/sessions/state/)
* `gemini.adk.state.user_scope` (agent.event.append) — [vendor docs](https://google.github.io/adk-docs/sessions/state/)
* `gemini.adk.streaming` (agent.adk\_streaming) — [vendor docs](https://google.github.io/adk-docs/streaming/)
* `gemini.adk.tools.agent_tool` (tool.call) — [vendor docs](https://google.github.io/adk-docs/tools/agent-tools/)
* `gemini.adk.tools.function_tool` (tool.call) — [vendor docs](https://google.github.io/adk-docs/tools/function-tools/)
* `gemini.adk.tools.long_running_function_tool` (tool.call) — [vendor docs](https://google.github.io/adk-docs/tools/function-tools/)
* `gemini.adk.tools.mcp_toolset` (tool.mcp) — [vendor docs](https://google.github.io/adk-docs/tools/mcp-tools/)
* `gemini.adk.tools.openapi` (tool.call) — [vendor docs](https://google.github.io/adk-docs/tools/openapi-tools/)
* `gemini.adk.tools.toolbox` (tool.call) — [vendor docs](https://google.github.io/adk-docs/tools/google-cloud-tools/)
* `gemini.adk.workflow_agents` (agent.adk\_workflow) — [vendor docs](https://google.github.io/adk-docs/agents/workflow-agents/)

**Parameters** (46):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                               |
| --------- | ------ | ------- | ------- | ---- | ----------------------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/multi-agents/)                      |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/custom-agents/)                     |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/llm-agents/)                        |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/workflow-agents/loop-agents/)       |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/workflow-agents/parallel-agents/)   |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/workflow-agents/sequential-agents/) |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/artifacts/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/callbacks/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/callbacks/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/callbacks/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/callbacks/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/callbacks/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/callbacks/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/callbacks/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/runtime/)                                  |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/evaluate/)                                 |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/get-started/quickstart/)                   |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/get-started/quickstart/)                   |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/evaluate/)                                 |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/evaluate/)                                 |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/events/)                                   |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/llm-agents/)                        |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/workflow-agents/loop-agents/)       |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/sessions/memory/)                          |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/multi-agents/)                      |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/llm-agents/)                        |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/llm-agents/)                        |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/llm-agents/)                        |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/multi-agents/)                      |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/multi-agents/)                      |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/runtime/)                                  |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/runtime/)                                  |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/sessions/session/)                         |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/sessions/session/)                         |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/sessions/state/)                           |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/sessions/state/)                           |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/sessions/state/)                           |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/sessions/state/)                           |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/streaming/)                                |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/tools/agent-tools/)                        |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/tools/function-tools/)                     |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/tools/function-tools/)                     |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/tools/mcp-tools/)                          |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/tools/openapi-tools/)                      |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/tools/google-cloud-tools/)                 |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/agents/workflow-agents/)                   |

## Antigravity Auth

**Capabilities** (4/4 rows usable):

* `gemini.antigravity.auth.gemini_api_key_env` (provider.auth.api\_key) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.auth.gemini_config_api_key` (provider.auth.api\_key) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.auth.harness_path_env` (provider.binary.override) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.auth.vertex` (provider.auth.vertex) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (4):

| Parameter            | Type    | Default | Allowed | Risk   | Notes                                                                |
| -------------------- | ------- | ------- | ------- | ------ | -------------------------------------------------------------------- |
| `env.GEMINI_API_KEY` | string  | —       | —       | high   | Secret-bearing env var; operator-only, never a chat-surface knob.    |
| `option`             | object  | —       | —       | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`             | object  | —       | —       | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `vertex`             | boolean | `false` | —       | medium | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity Built-in Tools

> Concept guide: [Tools & MCP](/concepts/tools-and-mcp) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (8/12 rows usable):

* `gemini.antigravity.tool.custom_python` (tool.custom.python) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.tool.edit_file` (tool.fs.edit) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.tool.find_file` (tool.fs.find) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.tool.finish` (agent.finish) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.tool.list_directory` (tool.fs.list) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.tool.search_directory` (tool.fs.grep) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.tool.start_subagent` (agent.subagent.spawn) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.tool.view_file` (tool.fs.read) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (8):

| Parameter      | Type   | Default | Allowed | Risk | Notes                                                                |
| -------------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`       | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`       | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`       | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `tools.finish` | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`       | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`       | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`       | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`       | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity CLI

**Capabilities** (5/13 rows usable):

* `gemini.antigravity.cli.add_dir` (agent.workspace.scope) — [vendor docs](https://antigravity.google/docs/cli-getting-started)
* `gemini.antigravity.cli.conversation_resume` (agent.session.resume) — [vendor docs](https://antigravity.google/docs/cli-getting-started)
* `gemini.antigravity.cli.print` (agent.run.print) — [vendor docs](https://antigravity.google/docs/cli-getting-started)
* `gemini.antigravity.cli.print_timeout` (agent.run.print) — [vendor docs](https://antigravity.google/docs/cli-getting-started)
* `gemini.antigravity.cli.subcommand.help` (agent.help) — [vendor docs](https://antigravity.google/docs/cli-getting-started)

**Parameters** (8):

| Parameter              | Type    | Default | Allowed | Risk   | Notes                                                                  |
| ---------------------- | ------- | ------- | ------- | ------ | ---------------------------------------------------------------------- |
| `argv.--print`         | string  | —       | —       | low    | [docs](https://antigravity.google/docs/cli-getting-started)            |
| `option`               | object  | —       | —       | low    | [docs](https://antigravity.google/docs/cli-getting-started)            |
| `argv.--add-dir`       | array   | —       | —       | medium | [docs](https://antigravity.google/docs/cli-getting-started)            |
| `argv.--continue`      | boolean | —       | —       | low    | [docs](https://antigravity.google/docs/cli-getting-started)            |
| `argv.--conversation`  | string  | —       | —       | low    | SDK twin: LocalAgentConfig.conversation\_id (build\_agent\_config).    |
| `argv.--log-file`      | string  | —       | —       | low    | [docs](https://antigravity.google/docs/cli-getting-started)            |
| `argv.--print-timeout` | string  | —       | —       | low    | Vendor default (5m0s today) is the binary's, never restated here (T9). |
| `argv.--sandbox`       | boolean | —       | —       | medium | [docs](https://antigravity.google/docs/cli-getting-started)            |

## Antigravity Connections

**Capabilities** (1/2 rows usable):

* `gemini.antigravity.connection.local` (agent.connection.local) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (1):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity Distribution

**Capabilities** (2/6 rows usable):

* `gemini.antigravity.distribution.cli_install_script` (provider.distribution.install) — [vendor docs](https://antigravity.google/docs/cli-getting-started)
* `gemini.antigravity.distribution.sdk_pypi` (provider.distribution.pypi) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (2):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://antigravity.google/docs/cli-getting-started)          |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity Errors

**Capabilities** (3/3 rows usable):

* `gemini.antigravity.error.connection` (agent.error.connection) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.error.execution` (agent.error.execution) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.error.validation` (agent.error.validation) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (3):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity Hooks

> Concept guide: [Hooks](/concepts/hooks) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (7/9 rows usable):

* `gemini.antigravity.hook.on_session_end` (agent.hook.session\_end) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.hook.on_session_start` (agent.hook.session\_start) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.hook.on_tool_error` (agent.hook.tool\_error) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.hook.post_tool_call` (agent.hook.post\_tool) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.hook.post_turn` (agent.hook.post\_turn) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.hook.pre_tool_call_decide` (agent.hook.pre\_tool) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.hook.pre_turn` (agent.hook.pre\_turn) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (7):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity MCP

> Concept guide: [Tools & MCP](/concepts/tools-and-mcp) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (3/4 rows usable):

* `gemini.antigravity.mcp.bridge` (provider.mcp.bridge) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.mcp.http` (provider.mcp.client) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.mcp.stdio` (provider.mcp.client) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (3):

| Parameter     | Type   | Default | Allowed | Risk | Notes                                                                                                                                                        |
| ------------- | ------ | ------- | ------- | ---- | ------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `option`      | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                         |
| `option`      | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                         |
| `mcp_servers` | object | —       | —       | high | Transport discriminators resolved from the installed SDK union (stdio/http today; the historic sse transport was folded into streamable-http by the vendor). |

## Antigravity Models

**Capabilities** (3/4 rows usable):

* `gemini.antigravity.model.gemini_2_5_pro` (llm.complete) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.model.gemini_3_1_pro_preview` (llm.complete) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.model.gemini_3_5_flash` (llm.complete) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (3):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity Multimodal

**Capabilities** (5/5 rows usable):

* `gemini.antigravity.input.audio` (media.audio.understand) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.input.document` (media.document.understand) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.input.from_file` (media.from\_file) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.input.image` (media.image.understand) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.input.video` (media.video.understand) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (5):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity Policies

**Capabilities** (7/8 rows usable):

* `gemini.antigravity.policy.allow` (agent.policy.allow) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.policy.ask_user` (agent.policy.ask\_user) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.policy.confirm_run_command` (agent.policy.confirm) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.policy.deny` (agent.policy.deny) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.policy.deny_all` (agent.policy.deny) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.policy.safe_defaults` (agent.policy.safe\_defaults) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.policy.workspace_only` (agent.policy.workspace\_scope) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (7):

| Parameter    | Type   | Default | Allowed | Risk   | Notes                                                                |
| ------------ | ------ | ------- | ------- | ------ | -------------------------------------------------------------------- |
| `workspaces` | array  | —       | —       | medium | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`     | object | —       | —       | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`     | object | —       | —       | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`     | object | —       | —       | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`     | object | —       | —       | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`     | object | —       | —       | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`     | object | —       | —       | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity SDK

**Capabilities** (12/12 rows usable):

* `gemini.antigravity.sdk.agent_class` (agent.lifecycle) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.agent_config` (agent.config) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.capabilities_config` (agent.capabilities) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.conversation` (agent.conversation.lifecycle) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.gemini_config` (provider.auth) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.generation_config` (llm.complete) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.local_agent_config` (agent.config) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.model_config` (llm.model.select) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.model_entry` (llm.model.select) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.thinking_level` (llm.thinking) — [vendor docs](https://ai.google.dev/gemini-api/docs/thinking)
* `gemini.antigravity.sdk.tool_context` (tool.context) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.sdk.usage_metadata` (agent.observability.usage) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (16):

| Parameter                                                | Type    | Default | Allowed                                    | Risk   | Notes                                                                                                                                                                                                                                                                                                                                |
| -------------------------------------------------------- | ------- | ------- | ------------------------------------------ | ------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `capabilities.compaction_threshold`                      | number  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `capabilities.disabled_tools`                            | array   | —       | —                                          | medium | BuiltinTools denylist; mutually exclusive with enabled\_tools (bridge fails loud). Live finding 2026-06-11: the localharness executed an edit\_file call despite a disabled\_tools=\[edit\_file] denylist and an enabled\_tools allowlist without it - treat filter enforcement as advisory (pc-ag-rows evidence toolswrite.json).   |
| `capabilities.enable_subagents`                          | boolean | `false` | —                                          | medium | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `capabilities.enabled_tools`                             | array   | —       | —                                          | medium | BuiltinTools allowlist; mutually exclusive with disabled\_tools (bridge fails loud). Live finding 2026-06-11: the localharness executed an edit\_file call despite a disabled\_tools=\[edit\_file] denylist and an enabled\_tools allowlist without it - treat filter enforcement as advisory (pc-ag-rows evidence toolswrite.json). |
| `capabilities.image_model`                               | string  | —       | —                                          | low    | Default resolved from the installed SDK (DEFAULT\_IMAGE\_GENERATION\_MODEL), never pinned here.                                                                                                                                                                                                                                      |
| `model`                                                  | string  | —       | —                                          | low    | Default resolved from the installed SDK (types.DEFAULT\_MODEL), never pinned here.                                                                                                                                                                                                                                                   |
| `gemini_config.models.default.generation.thinking_level` | enum    | —       | `"minimal"`, `"low"`, `"medium"`, `"high"` | low    | [docs](https://ai.google.dev/gemini-api/docs/thinking)                                                                                                                                                                                                                                                                               |
| `agent`                                                  | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `option`                                                 | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `option`                                                 | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `option`                                                 | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `option`                                                 | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `local_agent_config`                                     | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `option`                                                 | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `option`                                                 | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |
| `option`                                                 | object  | —       | —                                          | low    | [docs](https://github.com/google-antigravity/antigravity-sdk-python)                                                                                                                                                                                                                                                                 |

## Antigravity Streaming

**Capabilities** (5/5 rows usable):

* `gemini.antigravity.stream.chat_response` (llm.stream) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.stream.structured_output` (llm.structured\_output) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.stream.text_deltas` (llm.stream) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.stream.thoughts` (llm.thinking) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.stream.tool_calls` (tool.call) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (5):

| Parameter                              | Type    | Default | Allowed | Risk | Notes                                                                       |
| -------------------------------------- | ------- | ------- | ------- | ---- | --------------------------------------------------------------------------- |
| `capabilities.finish_tool_schema_json` | string  | —       | —       | low  | JSON schema for the finish tool — drives ChatResponse.structured\_output(). |
| `stream`                               | boolean | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python)        |
| `option`                               | object  | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python)        |
| `option`                               | object  | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python)        |
| `option`                               | object  | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python)        |

## Antigravity System Instructions

**Capabilities** (2/2 rows usable):

* `gemini.antigravity.system.custom` (agent.system\_instructions) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.system.templated` (agent.system\_instructions) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (2):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Antigravity Triggers

**Capabilities** (4/5 rows usable):

* `gemini.antigravity.trigger.context` (agent.trigger.context) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.trigger.decorator` (agent.trigger.decorator) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.trigger.every` (agent.trigger.interval) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)
* `gemini.antigravity.trigger.on_file_change` (agent.trigger.file\_watch) — [vendor docs](https://github.com/google-antigravity/antigravity-sdk-python)

**Parameters** (4):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |
| `option`  | object | —       | —       | low  | [docs](https://github.com/google-antigravity/antigravity-sdk-python) |

## Audio generation

> Concept guide: [Voice & Realtime](/concepts/voice-and-realtime) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (18/21 rows usable):

* `gemini.chirp.chirp_3` (media.audio.tts) — [vendor docs](https://cloud.google.com/text-to-speech/docs/chirp3-hd)
* `gemini.lyria.generate_music` (media.music.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/music-generation)
* `gemini.lyria.realtime` (realtime.music) — [vendor docs](https://ai.google.dev/api/live_music)
* `gemini.tts` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.tts.gemini_25_flash` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.tts.gemini_25_pro` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.tts.language_code` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.tts.multispeaker` (media.audio.tts\_multispeaker) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.tts.params.audio_encoding.linear16` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.tts.params.audio_encoding.mp3` (media.audio.tts) — [vendor docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig)
* `gemini.tts.params.audio_encoding.ogg_opus` (media.audio.tts) — [vendor docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig)
* `gemini.tts.params.effects_profile` (media.audio.tts) — [vendor docs](https://cloud.google.com/text-to-speech/docs/audio-profiles)
* `gemini.tts.params.pitch` (media.audio.tts) — [vendor docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig)
* `gemini.tts.params.sample_rate` (media.audio.tts) — [vendor docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig)
* `gemini.tts.params.speaking_rate` (media.audio.tts) — [vendor docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig)
* `gemini.tts.params.volume_gain` (media.audio.tts) — [vendor docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig)
* `gemini.tts.style_control` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.tts.voice.prebuilt` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)

**Parameters** (18):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                              |
| --------- | ------ | ------- | ------- | ---- | ---------------------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/text-to-speech/docs/chirp3-hd)                     |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/music-generation)                     |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/live_music)                                       |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                    |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                    |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                    |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                    |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                    |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                    |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig) |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig) |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/text-to-speech/docs/audio-profiles)                |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig) |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig) |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig) |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/text-to-speech/docs/reference/rest/v1/AudioConfig) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                    |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                    |

## Auth

**Capabilities** (1/4 rows usable):

* `gemini.auth.api_key` (provider.admin.read) — [vendor docs](https://ai.google.dev/gemini-api/docs/api-key)

**Parameters** (1):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                 |
| --------- | ------ | ------- | ------- | ---- | ----------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/api-key) |

## Batch

> Concept guide: [Cost, Batch & Cache](/concepts/cost-batch-cache) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (8/8 rows usable):

* `gemini.batch` (batch.run) — [vendor docs](https://ai.google.dev/gemini-api/docs/batch-mode)
* `gemini.batches.cancel` (batch.run) — [vendor docs](https://ai.google.dev/api/batches)
* `gemini.batches.create` (batch.run) — [vendor docs](https://ai.google.dev/gemini-api/docs/batch-mode)
* `gemini.batches.delete` (batch.run) — [vendor docs](https://ai.google.dev/api/batches)
* `gemini.batches.file_input` (batch.run) — [vendor docs](https://ai.google.dev/gemini-api/docs/batch-mode)
* `gemini.batches.get` (batch.run) — [vendor docs](https://ai.google.dev/api/batches)
* `gemini.batches.inline_input` (batch.run) — [vendor docs](https://ai.google.dev/gemini-api/docs/batch-mode)
* `gemini.batches.list` (batch.run) — [vendor docs](https://ai.google.dev/api/batches)

**Parameters** (8):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                    |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/batch-mode) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/batches)                |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/batch-mode) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/batches)                |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/batch-mode) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/batches)                |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/batch-mode) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/batches)                |

## Caching

> Concept guide: [Cost, Batch & Cache](/concepts/cost-batch-cache) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (10/10 rows usable):

* `gemini.caches.create` (llm.cache.create) — [vendor docs](https://ai.google.dev/api/caching)
* `gemini.caches.delete` (llm.cache.delete) — [vendor docs](https://ai.google.dev/api/caching)
* `gemini.caches.get` (llm.cache.get) — [vendor docs](https://ai.google.dev/api/caching)
* `gemini.caches.list` (llm.cache.list) — [vendor docs](https://ai.google.dev/api/caching)
* `gemini.caches.system_instruction` (llm.cache.create) — [vendor docs](https://ai.google.dev/gemini-api/docs/caching)
* `gemini.caches.tools` (llm.cache.create) — [vendor docs](https://ai.google.dev/gemini-api/docs/caching)
* `gemini.caches.ttl` (llm.cache.update) — [vendor docs](https://ai.google.dev/gemini-api/docs/caching)
* `gemini.caches.update` (llm.cache.update) — [vendor docs](https://ai.google.dev/api/caching)
* `gemini.context_caching` (llm.complete) · [*partial-compatible*](#how-to-read-this-page) — [vendor docs](https://ai.google.dev/gemini-api/docs/caching)
* `gemini.context_caching.implicit` (llm.implicit\_cache) — [vendor docs](https://ai.google.dev/gemini-api/docs/caching)

**Parameters** (9):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                 |
| --------- | ------ | ------- | ------- | ---- | ----------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/caching)             |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/caching)             |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/caching)             |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/caching)             |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/caching) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/caching) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/caching) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/caching)             |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/caching) |

## Chats

**Capabilities** (1/1 rows usable):

* `gemini.chats.create` (llm.chats\_create) — [vendor docs](https://ai.google.dev/gemini-api/docs/text-generation)

**Parameters** (1):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                         |
| --------- | ------ | ------- | ------- | ---- | ------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/text-generation) |

## Core generation

**Capabilities** (8/8 rows usable):

* `gemini.flex_inference` (llm.flex\_inference) — [vendor docs](https://ai.google.dev/gemini-api/docs)
* `gemini.generate_content` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/text-generation)
* `gemini.generate_content.multi_turn` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/text-generation#chat)
* `gemini.generate_content.stream` (llm.stream) — [vendor docs](https://ai.google.dev/gemini-api/docs/text-generation#streaming-responses)
* `gemini.generate_content.system_instruction` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/text-generation#system-instructions)
* `gemini.interactions` (llm.interactions) — [vendor docs](https://ai.google.dev/api/interactions-api)
* `gemini.openai_compat` (llm.complete) · [*partial-compatible*](#how-to-read-this-page) — [vendor docs](https://ai.google.dev/gemini-api/docs/openai)
* `gemini.priority_inference` (llm.priority\_inference) — [vendor docs](https://ai.google.dev/gemini-api/docs)

**Parameters** (9):

| Parameter                                        | Type   | Default              | Allowed | Risk   | Notes                                                                                |
| ------------------------------------------------ | ------ | -------------------- | ------- | ------ | ------------------------------------------------------------------------------------ |
| `option`                                         | object | —                    | —       | low    | [docs](https://ai.google.dev/gemini-api/docs)                                        |
| `generationConfig.maxOutputTokens`               | number | `1024`               | —       | medium | [docs](https://ai.google.dev/api/generate-content)                                   |
| `model`                                          | string | `"gemini-2.5-flash"` | —       | low    | [docs](https://ai.google.dev/api/generate-content)                                   |
| `option`                                         | object | —                    | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/text-generation#chat)                   |
| `generationConfig.thinkingConfig.thinkingBudget` | number | `0`                  | —       | medium | Setting >0 means thoughts will consume part of maxOutputTokens before text emission. |
| `systemInstruction`                              | string | —                    | —       | low    | deeda passes the compiled system channel; REST wraps it as a Content parts object.   |
| `generationConfig.temperature`                   | number | `1`                  | —       | low    | [docs](https://ai.google.dev/api/generate-content)                                   |
| `option`                                         | object | —                    | —       | low    | [docs](https://ai.google.dev/api/interactions-api)                                   |
| `option`                                         | object | —                    | —       | low    | [docs](https://ai.google.dev/gemini-api/docs)                                        |

## Embeddings

**Capabilities** (12/13 rows usable):

* `gemini.embeddings` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings)
* `gemini.embeddings.batch` (batch.embeddings) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings)
* `gemini.embeddings.output_dimensionality` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings)
* `gemini.embeddings.task_type.classification` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types)
* `gemini.embeddings.task_type.clustering` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types)
* `gemini.embeddings.task_type.code_retrieval_query` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types)
* `gemini.embeddings.task_type.fact_verification` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types)
* `gemini.embeddings.task_type.question_answering` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types)
* `gemini.embeddings.task_type.retrieval_document` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types)
* `gemini.embeddings.task_type.retrieval_query` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types)
* `gemini.embeddings.task_type.semantic_similarity` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types)
* `gemini.embeddings.title` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings)

**Parameters** (13):

| Parameter  | Type   | Default                  | Allowed                                                                                                                                                                           | Risk | Notes                                                               |
| ---------- | ------ | ------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---- | ------------------------------------------------------------------- |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings)            |
| `model`    | string | `"gemini-embedding-001"` | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings)            |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings)            |
| `taskType` | enum   | `"SEMANTIC_SIMILARITY"`  | `"SEMANTIC_SIMILARITY"`, `"CLASSIFICATION"`, `"CLUSTERING"`, `"RETRIEVAL_DOCUMENT"`, `"RETRIEVAL_QUERY"`, `"QUESTION_ANSWERING"`, `"FACT_VERIFICATION"`, `"CODE_RETRIEVAL_QUERY"` | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings)            |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types) |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types) |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types) |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types) |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types) |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types) |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types) |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings#task-types) |
| `option`   | object | —                        | —                                                                                                                                                                                 | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings)            |

## File search

**Capabilities** (8/9 rows usable):

* `gemini.file_search.documents.get` (vector\_store.documents.get) — [vendor docs](https://ai.google.dev/api/file-search/documents)
* `gemini.file_search.documents.list` (vector\_store.documents.list) — [vendor docs](https://ai.google.dev/api/file-search/documents)
* `gemini.file_search.stores.create` (vector\_store.create) — [vendor docs](https://ai.google.dev/api/file-search/file-search-stores)
* `gemini.file_search.stores.delete` (vector\_store.delete) — [vendor docs](https://ai.google.dev/api/file-search/file-search-stores)
* `gemini.file_search.stores.get` (vector\_store.get) — [vendor docs](https://ai.google.dev/api/file-search/file-search-stores)
* `gemini.file_search.stores.import` (vector\_store.import) — [vendor docs](https://ai.google.dev/api/file-search/file-search-stores)
* `gemini.file_search.stores.list` (vector\_store.list) — [vendor docs](https://ai.google.dev/api/file-search/file-search-stores)
* `gemini.file_search.stores.upload` (vector\_store.upload) — [vendor docs](https://ai.google.dev/api/file-search/file-search-stores)

**Parameters** (8):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                            |
| --------- | ------ | ------- | ------- | ---- | ---------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/file-search/documents)          |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/file-search/documents)          |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/file-search/file-search-stores) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/file-search/file-search-stores) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/file-search/file-search-stores) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/file-search/file-search-stores) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/file-search/file-search-stores) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/file-search/file-search-stores) |

## Files

**Capabilities** (6/6 rows usable):

* `gemini.files.delete` (file.delete) — [vendor docs](https://ai.google.dev/api/files)
* `gemini.files.get` (file.get) — [vendor docs](https://ai.google.dev/api/files)
* `gemini.files.list` (file.list) — [vendor docs](https://ai.google.dev/api/files)
* `gemini.files.media_upload` (file.upload) — [vendor docs](https://ai.google.dev/api/files#method:-media.upload)
* `gemini.files.upload` (file.upload) — [vendor docs](https://ai.google.dev/gemini-api/docs/files)
* `gemini.files.uri_reference` (file.get) — [vendor docs](https://ai.google.dev/gemini-api/docs/files)

**Parameters** (6):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                        |
| --------- | ------ | ------- | ------- | ---- | ------------------------------------------------------------ |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/files)                      |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/files)                      |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/files)                      |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/files#method:-media.upload) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/files)          |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/files)          |

## Function calling

**Capabilities** (10/11 rows usable):

* `gemini.function_calling` (tool.call) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling)
* `gemini.function_calling.allowed_function_names` (tool.call) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling#function-calling-modes)
* `gemini.function_calling.compositional` (tool.compositional\_calls) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling)
* `gemini.function_calling.declarations` (tool.call) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling)
* `gemini.function_calling.mcp` (tool.mcp) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling)
* `gemini.function_calling.mode.any` (tool.call) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling#function-calling-modes)
* `gemini.function_calling.mode.auto` (tool.call) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling#function-calling-modes)
* `gemini.function_calling.mode.none` (tool.call) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling#function-calling-modes)
* `gemini.function_calling.parallel` (tool.parallel\_calls) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling)
* `gemini.function_calling.response_part` (tool.call) — [vendor docs](https://ai.google.dev/gemini-api/docs/function-calling)

**Parameters** (10):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                                 |
| --------- | ------ | ------- | ------- | ---- | ------------------------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling#function-calling-modes) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling)                        |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling)                        |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling)                        |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling#function-calling-modes) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling#function-calling-modes) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling#function-calling-modes) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling)                        |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling)                        |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/function-calling)                        |

## generationConfig

**Capabilities** (16/22 rows usable):

* `gemini.generation_config.candidate_count` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)
* `gemini.generation_config.max_output_tokens` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)
* `gemini.generation_config.media_resolution` (media.image.understand) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)
* `gemini.generation_config.response_mime_type` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/structured-output)
* `gemini.generation_config.response_modalities.audio` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.generation_config.response_modalities.image` (media.image.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/image-generation)
* `gemini.generation_config.response_modalities.text` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)
* `gemini.generation_config.response_schema` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/structured-output)
* `gemini.generation_config.seed` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)
* `gemini.generation_config.speech_config` (media.audio.tts) — [vendor docs](https://ai.google.dev/gemini-api/docs/speech-generation)
* `gemini.generation_config.stop_sequences` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)
* `gemini.generation_config.temperature` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)
* `gemini.generation_config.thinking_config.include_thoughts` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/thinking)
* `gemini.generation_config.thinking_config.thinking_budget` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/thinking)
* `gemini.generation_config.top_k` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)
* `gemini.generation_config.top_p` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-content#generationconfig)

**Parameters** (16):

| Parameter                                         | Type    | Default | Allowed         | Risk   | Notes                                                                                |
| ------------------------------------------------- | ------- | ------- | --------------- | ------ | ------------------------------------------------------------------------------------ |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/api/generate-content#generationconfig)                  |
| `generationConfig.maxOutputTokens`                | number  | `1024`  | —               | medium | Hard cap on tokens emitted; medium risk because misconfiguration burns cost.         |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/api/generate-content#generationconfig)                  |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/gemini-api/docs/structured-output)                      |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                      |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/gemini-api/docs/image-generation)                       |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/api/generate-content#generationconfig)                  |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/gemini-api/docs/structured-output)                      |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/api/generate-content#generationconfig)                  |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/gemini-api/docs/speech-generation)                      |
| `option`                                          | object  | —       | —               | low    | [docs](https://ai.google.dev/api/generate-content#generationconfig)                  |
| `generationConfig.temperature`                    | number  | `1`     | —               | low    | [docs](https://ai.google.dev/api/generate-content#generationconfig)                  |
| `generationConfig.thinkingConfig.includeThoughts` | boolean | `false` | `true`, `false` | low    | [docs](https://ai.google.dev/gemini-api/docs/thinking)                               |
| `generationConfig.thinkingConfig.thinkingBudget`  | number  | —       | —               | medium | Thoughts consume part of maxOutputTokens before text emission on thinking-on models. |
| `generationConfig.topK`                           | number  | —       | —               | low    | [docs](https://ai.google.dev/api/generate-content#generationconfig)                  |
| `generationConfig.topP`                           | number  | —       | —               | low    | [docs](https://ai.google.dev/api/generate-content#generationconfig)                  |

## Grounding

**Capabilities** (6/8 rows usable):

* `gemini.grounding.google_maps` (tool.maps) — [vendor docs](https://ai.google.dev/gemini-api/docs/maps-grounding)
* `gemini.grounding.google_search` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/grounding)
* `gemini.grounding.grounding_chunks` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/grounding)
* `gemini.grounding.grounding_supports` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/grounding)
* `gemini.grounding.search_entry_point` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/grounding)
* `gemini.grounding.web_search_queries` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/grounding)

**Parameters** (6):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                        |
| --------- | ------ | ------- | ------- | ---- | ------------------------------------------------------------ |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/maps-grounding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/grounding)      |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/grounding)      |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/grounding)      |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/grounding)      |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/grounding)      |

## Image generation

**Capabilities** (13/26 rows usable):

* `gemini.image_generation` (media.image.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/image-generation)
* `gemini.image_generation.edit` (media.image.edit) — [vendor docs](https://ai.google.dev/gemini-api/docs/image-generation)
* `gemini.imagen.aspect_ratio` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.imagen_4` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.imagen_4_fast` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.imagen_4_ultra` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.params.compression_quality` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.params.guidance_scale` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.params.output_format` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.params.sample_count` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.person_generation` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.predict` (media.image.predict) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)
* `gemini.imagen.safety_filter_level` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/imagen)

**Parameters** (13):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                          |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/image-generation) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/image-generation) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/imagen)           |

## Live API

**Capabilities** (30/34 rows usable):

* `gemini.live_api` (realtime.session.create) — [vendor docs](https://ai.google.dev/gemini-api/docs/live)
* `gemini.live.affective_dialog` (realtime.affective\_dialog) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.audio_transcription` (realtime.transcription) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.barge_in` (realtime.barge\_in) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.context_window_compression` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.ephemeral_tokens` (realtime.ephemeral\_token) — [vendor docs](https://ai.google.dev/gemini-api/docs/ephemeral-tokens)
* `gemini.live.event.bidi_client_content` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#bidigeneratecontentclientcontent)
* `gemini.live.event.bidi_realtime_input` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#bidigeneratecontentrealtimeinput)
* `gemini.live.event.bidi_server_content` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#bidigeneratecontentservercontent)
* `gemini.live.event.bidi_setup` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#bidigeneratecontentsetup)
* `gemini.live.event.bidi_setup_complete` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#bidigeneratecontentsetupcomplete)
* `gemini.live.event.bidi_tool_call` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#bidigeneratecontenttoolcall)
* `gemini.live.event.bidi_tool_response` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#bidigeneratecontenttoolresponse)
* `gemini.live.event.session_resumption_update` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#sessionresumptionupdate)
* `gemini.live.event.usage_metadata` (realtime.session) — [vendor docs](https://ai.google.dev/api/live#usagemetadata)
* `gemini.live.input.audio` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.input.screen` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.input.video` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.interrupted_event` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.language_code` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.modality.audio` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.modality.text` (realtime.session) · [*model-dependent*](#how-to-read-this-page) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.proactive_audio` (realtime.proactive\_audio) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.session_management` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-api/session-management)
* `gemini.live.tools.function_calling` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.turn_complete` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.vad.automatic` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.vad.manual` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.voice_config` (realtime.session) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-guide)
* `gemini.live.websocket` (realtime.websocket) — [vendor docs](https://ai.google.dev/gemini-api/docs/live-api/get-started-websocket)

**Parameters** (29):

| Parameter                   | Type   | Default     | Allowed | Risk   | Notes                                                                        |
| --------------------------- | ------ | ----------- | ------- | ------ | ---------------------------------------------------------------------------- |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live)                           |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/ephemeral-tokens)               |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#bidigeneratecontentclientcontent)      |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#bidigeneratecontentrealtimeinput)      |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#bidigeneratecontentservercontent)      |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#bidigeneratecontentsetupcomplete)      |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#bidigeneratecontentsetup)              |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#bidigeneratecontenttoolcall)           |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#bidigeneratecontenttoolresponse)       |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#sessionresumptionupdate)               |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/api/live#usagemetadata)                         |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `config.responseModalities` | array  | `["AUDIO"]` | —       | medium | [docs](https://ai.google.dev/gemini-api/docs/live)                           |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-guide)                     |
| `option`                    | object | —           | —       | low    | [docs](https://ai.google.dev/gemini-api/docs/live-api/get-started-websocket) |

## MCP

> Concept guide: [Tools & MCP](/concepts/tools-and-mcp) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (2/6 rows usable):

* `gemini.mcp.adk_toolset_client` (tool.mcp) — [vendor docs](https://google.github.io/adk-docs/tools/mcp-tools/)
* `gemini.mcp.toolbox_databases` (tool.mcp) — [vendor docs](https://googleapis.github.io/genai-toolbox/)

**Parameters** (2):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                      |
| --------- | ------ | ------- | ------- | ---- | ---------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://google.github.io/adk-docs/tools/mcp-tools/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/genai-toolbox/)        |

## MCP Toolbox

> Concept guide: [Tools & MCP](/concepts/tools-and-mcp) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (3/14 rows usable):

* `gemini.mcp.toolbox.http` (tool.mcp) — [vendor docs](https://googleapis.github.io/genai-toolbox/resources/sources/http/)
* `gemini.mcp.toolbox.parameterized_sql` (tool.mcp) — [vendor docs](https://googleapis.github.io/genai-toolbox/resources/tools/)
* `gemini.mcp.toolbox.sqlite` (tool.mcp) — [vendor docs](https://googleapis.github.io/genai-toolbox/resources/sources/sqlite/)

**Parameters** (3):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                        |
| --------- | ------ | ------- | ------- | ---- | ---------------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/genai-toolbox/resources/sources/http/)   |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/genai-toolbox/resources/tools/)          |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/genai-toolbox/resources/sources/sqlite/) |

## MIME types

**Capabilities** (18/18 rows usable):

* `gemini.mime.application.pdf` (media.document.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/document-processing)
* `gemini.mime.audio.aac` (media.audio.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/audio)
* `gemini.mime.audio.aiff` (media.audio.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/audio)
* `gemini.mime.audio.flac` (media.audio.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/audio)
* `gemini.mime.audio.mp3` (media.audio.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/audio)
* `gemini.mime.audio.ogg` (media.audio.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/audio)
* `gemini.mime.audio.wav` (media.audio.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/audio)
* `gemini.mime.image.heic` (media.image.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/vision)
* `gemini.mime.image.heif` (media.image.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/vision)
* `gemini.mime.image.jpeg` (media.image.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/vision)
* `gemini.mime.image.png` (media.image.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/vision)
* `gemini.mime.image.webp` (media.image.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/vision)
* `gemini.mime.video.avi` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)
* `gemini.mime.video.mov` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)
* `gemini.mime.video.mp4` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)
* `gemini.mime.video.mpeg` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)
* `gemini.mime.video.threegpp` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)
* `gemini.mime.video.webm` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)

**Parameters** (18):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                             |
| --------- | ------ | ------- | ------- | ---- | ----------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/document-processing) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/audio)               |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/audio)               |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/audio)               |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/audio)               |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/audio)               |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/audio)               |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/vision)              |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/vision)              |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/vision)              |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/vision)              |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/vision)              |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |

## Models

**Capabilities** (13/43 rows usable):

* `gemini.model.context_window.1m` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/long-context)
* `gemini.model.gemini_25_flash` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/models)
* `gemini.model.gemini_25_flash_image` (media.image.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/image-generation)
* `gemini.model.gemini_25_flash_lite` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/models)
* `gemini.model.gemini_25_pro` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/models)
* `gemini.model.gemini_3_flash` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/models)
* `gemini.model.gemini_3_pro_image` (media.image.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/image-generation)
* `gemini.model.gemini_embedding_001` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings)
* `gemini.model.gemini_embedding_2` (llm.embed) — [vendor docs](https://ai.google.dev/gemini-api/docs/embeddings)
* `gemini.models.get` (provider.models\_list) — [vendor docs](https://ai.google.dev/api/models#method:-models.get)
* `gemini.models.list` (provider.models\_list) — [vendor docs](https://ai.google.dev/api/models)
* `gemini.models.predict` (llm.complete) — [vendor docs](https://ai.google.dev/api/models#method:-models.predict)
* `gemini.models.predict_long_running` (provider.long\_running\_operations) — [vendor docs](https://ai.google.dev/api/models#method:-models.predictLongRunning)

**Parameters** (13):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                      |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/long-context)                 |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/image-generation)             |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/models)                       |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/models)                       |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/models)                       |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/models)                       |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/image-generation)             |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings)                   |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/embeddings)                   |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/models#method:-models.get)                |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/models)                                   |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/models#method:-models.predictLongRunning) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/models#method:-models.predict)            |

## Multimodal

**Capabilities** (9/9 rows usable):

* `gemini.multimodal.audio_understanding` (media.audio.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/audio)
* `gemini.multimodal.audio.timestamp` (media.audio.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/audio)
* `gemini.multimodal.document_understanding` (media.document.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/document-processing)
* `gemini.multimodal.image_understanding` (media.image.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/vision)
* `gemini.multimodal.inline_data` (media.image.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/vision)
* `gemini.multimodal.video_understanding` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)
* `gemini.multimodal.video.clip` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)
* `gemini.multimodal.video.fps` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)
* `gemini.multimodal.video.youtube_url` (media.video.understand) — [vendor docs](https://ai.google.dev/gemini-api/docs/video-understanding)

**Parameters** (9):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                             |
| --------- | ------ | ------- | ------- | ---- | ----------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/audio)               |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/audio)               |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/document-processing) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/vision)              |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/vision)              |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video-understanding) |

## Observability

**Capabilities** (5/10 rows usable):

* `gemini.observability.audit_logs` (provider.admin.read) — [vendor docs](https://cloud.google.com/vertex-ai/docs/general/audit-logging)
* `gemini.observability.cloud_logging` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/logging/docs)
* `gemini.observability.cloud_monitoring` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/monitoring/docs)
* `gemini.observability.cloud_trace` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/trace/docs)
* `gemini.observability.quota_metrics` (provider.admin.read) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/quotas)

**Parameters** (5):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                 |
| --------- | ------ | ------- | ------- | ---- | --------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/vertex-ai/docs/general/audit-logging) |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/logging/docs)                         |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/monitoring/docs)                      |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/trace/docs)                           |
| `option`  | object | —       | —       | low  | [docs](https://cloud.google.com/vertex-ai/generative-ai/docs/quotas)  |

## Operations

**Capabilities** (3/5 rows usable):

* `gemini.operations.get` (provider.long\_running\_operations) — [vendor docs](https://ai.google.dev/api/operations#method:-operations.get)
* `gemini.operations.list` (provider.long\_running\_operations) — [vendor docs](https://ai.google.dev/api/operations#method:-operations.list)
* `gemini.operations.poll` (provider.long\_running\_operations) — [vendor docs](https://ai.google.dev/api/operations)

**Parameters** (3):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                                |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/operations#method:-operations.get)  |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/operations#method:-operations.list) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/operations)                         |

## Reasoning/thinking

> Concept guide: [Thinking & Reasoning](/concepts/thinking) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (4/5 rows usable):

* `gemini.thinking` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/thinking)
* `gemini.thinking.disabled` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/thinking)
* `gemini.thinking.dynamic` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/thinking)
* `gemini.thought_signatures` (llm.thought\_signatures) — [vendor docs](https://ai.google.dev/gemini-api/docs/thought-signatures)

**Parameters** (4):

| Parameter                                       | Type   | Default    | Allowed                                | Risk   | Notes                                                            |
| ----------------------------------------------- | ------ | ---------- | -------------------------------------- | ------ | ---------------------------------------------------------------- |
| `option`                                        | object | —          | —                                      | low    | [docs](https://ai.google.dev/gemini-api/docs/thinking)           |
| `option`                                        | object | —          | —                                      | low    | [docs](https://ai.google.dev/gemini-api/docs/thinking)           |
| `generationConfig.thinkingConfig.thinkingLevel` | enum   | `"medium"` | `"off"`, `"low"`, `"medium"`, `"high"` | medium | [docs](https://ai.google.dev/gemini-api/docs/thinking)           |
| `option`                                        | object | —          | —                                      | low    | [docs](https://ai.google.dev/gemini-api/docs/thought-signatures) |

## Runtime config

**Capabilities** (9/18 rows usable):

* `gemini.runtime.dual_mode.developer_api` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.runtime.dual_mode.vertex_ai` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.runtime.global_endpoint` (provider.admin.read) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/learn/locations#global-endpoint)
* `gemini.runtime.http.api_version` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.runtime.http.base_url` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.runtime.http.headers` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.runtime.http.timeout` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.runtime.location` (provider.admin.read) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/learn/locations)
* `gemini.runtime.project_id` (provider.admin.read) — [vendor docs](https://cloud.google.com/vertex-ai/docs/start/cloud-environment)

**Parameters** (5):

| Parameter | Type   | Default | Allowed | Risk | Notes                                              |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |

## Safety

**Capabilities** (7/9 rows usable):

* `gemini.safety_settings.civic_integrity` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/safety-settings)
* `gemini.safety_settings.dangerous_content` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/safety-settings)
* `gemini.safety_settings.harassment` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/safety-settings)
* `gemini.safety_settings.hate_speech` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/safety-settings)
* `gemini.safety_settings.sexually_explicit` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/safety-settings)
* `gemini.safety_settings.threshold.block_none` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/safety-settings)
* `gemini.safety_settings.threshold.block_only_high` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/safety-settings)

**Parameters** (7):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                         |
| --------- | ------ | ------- | ------- | ---- | ------------------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/safety-settings) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/safety-settings) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/safety-settings) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/safety-settings) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/safety-settings) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/safety-settings) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/safety-settings) |

## SDK JavaScript

**Capabilities** (8/12 rows usable):

* `gemini.sdk.js.batches` (batch.run) — [vendor docs](https://googleapis.github.io/js-genai/)
* `gemini.sdk.js.caches` (llm.cache.create) — [vendor docs](https://googleapis.github.io/js-genai/)
* `gemini.sdk.js.chats` (llm.chats\_create) — [vendor docs](https://googleapis.github.io/js-genai/)
* `gemini.sdk.js.files` (file.upload) — [vendor docs](https://googleapis.github.io/js-genai/)
* `gemini.sdk.js.google_genai` (llm.complete) — [vendor docs](https://googleapis.github.io/js-genai/)
* `gemini.sdk.js.live` (realtime.session.create) — [vendor docs](https://googleapis.github.io/js-genai/)
* `gemini.sdk.js.models` (llm.complete) — [vendor docs](https://googleapis.github.io/js-genai/)
* `gemini.sdk.js.operations` (provider.long\_running\_operations) — [vendor docs](https://googleapis.github.io/js-genai/)

**Parameters** (8):

| Parameter | Type   | Default | Allowed | Risk | Notes                                          |
| --------- | ------ | ------- | ------- | ---- | ---------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/js-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/js-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/js-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/js-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/js-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/js-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/js-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/js-genai/) |

## SDK Python

**Capabilities** (15/17 rows usable):

* `gemini.sdk.python.async_chats` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.batches` (batch.run) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.caches` (llm.cache.create) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.chats.create` (llm.chats\_create) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.chats.get_history` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.chats.send_message` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.chats.send_message_stream` (llm.stream) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.client` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.client.aio` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.errors` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.files` (file.upload) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.live.connect` (realtime.session.create) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.models` (provider.models\_list) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.operations` (provider.long\_running\_operations) — [vendor docs](https://googleapis.github.io/python-genai/)
* `gemini.sdk.python.types` (llm.complete) — [vendor docs](https://googleapis.github.io/python-genai/)

**Parameters** (15):

| Parameter | Type   | Default | Allowed | Risk | Notes                                              |
| --------- | ------ | ------- | ------- | ---- | -------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |
| `option`  | object | —       | —       | low  | [docs](https://googleapis.github.io/python-genai/) |

## Semantic Retriever

**Capabilities** (5/24 rows usable):

* `gemini.corpora.create` (vector\_store.create) — [vendor docs](https://ai.google.dev/api/semantic-retrieval/corpora)
* `gemini.corpora.delete` (vector\_store.delete) — [vendor docs](https://ai.google.dev/api/semantic-retrieval/corpora)
* `gemini.corpora.get` (vector\_store.get) — [vendor docs](https://ai.google.dev/api/semantic-retrieval/corpora)
* `gemini.corpora.list` (vector\_store.list) — [vendor docs](https://ai.google.dev/api/semantic-retrieval/corpora)
* `gemini.generate_answer` (llm.complete) — [vendor docs](https://ai.google.dev/api/generate-answer)

**Parameters** (5):

| Parameter | Type   | Default | Allowed | Risk | Notes                                                        |
| --------- | ------ | ------- | ------- | ---- | ------------------------------------------------------------ |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/semantic-retrieval/corpora) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/semantic-retrieval/corpora) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/semantic-retrieval/corpora) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/semantic-retrieval/corpora) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/generate-answer)            |

## Structured output

> Concept guide: [Structured Output](/concepts/structured-output) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (3/3 rows usable):

* `gemini.structured_output` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/structured-output)
* `gemini.structured_output.enum` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/structured-output#enum)
* `gemini.structured_output.property_ordering` (llm.complete) — [vendor docs](https://ai.google.dev/gemini-api/docs/structured-output)

**Parameters** (3):

| Parameter                           | Type   | Default              | Allowed                               | Risk | Notes                                                                |
| ----------------------------------- | ------ | -------------------- | ------------------------------------- | ---- | -------------------------------------------------------------------- |
| `option`                            | object | —                    | —                                     | low  | [docs](https://ai.google.dev/gemini-api/docs/structured-output#enum) |
| `option`                            | object | —                    | —                                     | low  | [docs](https://ai.google.dev/gemini-api/docs/structured-output)      |
| `generationConfig.responseMimeType` | enum   | `"application/json"` | `"application/json"`, `"text/x.enum"` | low  | [docs](https://ai.google.dev/gemini-api/docs/structured-output)      |

## Token counting

**Capabilities** (4/4 rows usable):

* `gemini.tokens.compute_cached` (llm.count\_tokens) — [vendor docs](https://ai.google.dev/api/tokens)
* `gemini.tokens.count` (llm.count\_tokens) — [vendor docs](https://ai.google.dev/gemini-api/docs/tokens)
* `gemini.tokens.count_message` (llm.count\_tokens) — [vendor docs](https://ai.google.dev/api/tokens)
* `gemini.tokens.usage_metadata` (llm.count\_tokens) — [vendor docs](https://ai.google.dev/gemini-api/docs/tokens)

**Parameters** (4):

| Parameter | Type   | Default              | Allowed | Risk | Notes                                                |
| --------- | ------ | -------------------- | ------- | ---- | ---------------------------------------------------- |
| `option`  | object | —                    | —       | low  | [docs](https://ai.google.dev/api/tokens)             |
| `option`  | object | —                    | —       | low  | [docs](https://ai.google.dev/api/tokens)             |
| `model`   | string | `"gemini-2.5-flash"` | —       | low  | [docs](https://ai.google.dev/api/tokens)             |
| `option`  | object | —                    | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/tokens) |

## Tools

> Concept guide: [Tools & MCP](/concepts/tools-and-mcp) — what this is, when to use it, and how to express it in workflow markdown.

**Capabilities** (6/9 rows usable):

* `gemini.tools.code_execution` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/code-execution)
* `gemini.tools.computer_use` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/computer-use)
* `gemini.tools.file_search` (tool.file\_search) — [vendor docs](https://ai.google.dev/gemini-api/docs/file-search)
* `gemini.tools.google_maps` (tool.maps) — [vendor docs](https://ai.google.dev/gemini-api/docs/maps-grounding)
* `gemini.tools.google_search` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/grounding)
* `gemini.tools.url_context` (provider.tool.server) — [vendor docs](https://ai.google.dev/gemini-api/docs/url-context)

**Parameters** (6):

| Parameter              | Type   | Default        | Allowed        | Risk | Notes                                                        |
| ---------------------- | ------ | -------------- | -------------- | ---- | ------------------------------------------------------------ |
| `option`               | object | —              | —              | low  | [docs](https://ai.google.dev/gemini-api/docs/code-execution) |
| `option`               | object | —              | —              | low  | [docs](https://ai.google.dev/gemini-api/docs/computer-use)   |
| `option`               | object | —              | —              | low  | [docs](https://ai.google.dev/gemini-api/docs/file-search)    |
| `option`               | object | —              | —              | low  | [docs](https://ai.google.dev/gemini-api/docs/maps-grounding) |
| `tools[].googleSearch` | object | `&#123;&#125;` | `&#123;&#125;` | low  | [docs](https://ai.google.dev/gemini-api/docs/grounding)      |
| `option`               | object | —              | —              | low  | [docs](https://ai.google.dev/gemini-api/docs/url-context)    |

## Vertex Agent Engine

**Capabilities** (16/22 rows usable):

* `gemini.vertex.agent_engine.create` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/deploy)
* `gemini.vertex.agent_engine.delete` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1beta1/projects.locations.reasoningEngines/delete)
* `gemini.vertex.agent_engine.get` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1beta1/projects.locations.reasoningEngines/get)
* `gemini.vertex.agent_engine.list` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1beta1/projects.locations.reasoningEngines/list)
* `gemini.vertex.agent_engine.memories.create` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/memory-bank/overview)
* `gemini.vertex.agent_engine.memories.delete` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/memory-bank/overview)
* `gemini.vertex.agent_engine.memories.generate` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/memory-bank/overview)
* `gemini.vertex.agent_engine.memories.list` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/memory-bank/overview)
* `gemini.vertex.agent_engine.memories.retrieve` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/memory-bank/overview)
* `gemini.vertex.agent_engine.patch` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1beta1/projects.locations.reasoningEngines/patch)
* `gemini.vertex.agent_engine.sessions.append_event` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/sessions/manage-sessions)
* `gemini.vertex.agent_engine.sessions.create` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/sessions/manage-sessions)
* `gemini.vertex.agent_engine.sessions.delete` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/sessions/manage-sessions)
* `gemini.vertex.agent_engine.sessions.get` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/sessions/manage-sessions)
* `gemini.vertex.agent_engine.sessions.list` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/sessions/manage-sessions)
* `gemini.vertex.agent_engine.sessions.list_events` (agent.event.append) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/sessions/manage-sessions)

## Vertex AI

**Capabilities** (6/10 rows usable):

* `gemini.vertex` (llm.complete) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/overview)
* `gemini.vertex.agent_builder` (agent.agent\_builder) — [vendor docs](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/agent-builder/overview)
* `gemini.vertex.batch_prediction` (batch.vertex) — [vendor docs](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/multimodal/batch-prediction-gemini)
* `gemini.vertex.grounding_search` (tool.vertex\_search) — [vendor docs](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/grounding/grounding-with-vertex-ai-search)
* `gemini.vertex.model_garden` (provider.model\_garden) — [vendor docs](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/model-garden/explore-models)
* `gemini.vertex.safety_filters` (provider.safety\_filters) — [vendor docs](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/multimodal/configure-safety-filters)

## Vertex Discovery Engine

**Capabilities** (11/12 rows usable):

* `gemini.vertex.discovery.answer` (llm.complete) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.engines.servingConfigs/answer)
* `gemini.vertex.discovery.data_stores.create` (vector\_store.create) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.dataStores/create)
* `gemini.vertex.discovery.data_stores.delete` (vector\_store.delete) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.dataStores/delete)
* `gemini.vertex.discovery.data_stores.get` (vector\_store.get) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.dataStores/get)
* `gemini.vertex.discovery.data_stores.list` (vector\_store.list) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.dataStores/list)
* `gemini.vertex.discovery.documents.import` (vector\_store.import) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.dataStores.branches.documents/import)
* `gemini.vertex.discovery.engines.create` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.engines/create)
* `gemini.vertex.discovery.engines.delete` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.engines/delete)
* `gemini.vertex.discovery.engines.get` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.engines/get)
* `gemini.vertex.discovery.engines.list` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.engines/list)
* `gemini.vertex.discovery.search` (vector\_store.search) — [vendor docs](https://cloud.google.com/discovery-engine/docs/reference/rest/v1/projects.locations.collections.engines.servingConfigs/search)

## Vertex Evaluation

**Capabilities** (12/12 rows usable):

* `gemini.vertex.eval.computation` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/online-pipeline-services)
* `gemini.vertex.eval.metric.bleu` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.metric.coherence` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.metric.exact_match` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.metric.fluency` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.metric.groundedness` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.metric.helpfulness` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.metric.rouge` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.metric.safety` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.pairwise` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.eval.pointwise` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/models/determine-eval)
* `gemini.vertex.evaluation` (eval.vertex) — [vendor docs](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/models/evaluation-overview)

## Vertex Model Garden

**Capabilities** (3/10 rows usable):

* `gemini.vertex.partner.deepseek` (llm.complete) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/open-models)
* `gemini.vertex.partner.openai_gpt_oss` (llm.complete) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/open-models)
* `gemini.vertex.partner.qwen` (llm.complete) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models)

## Vertex Platform

**Capabilities** (5/25 rows usable):

* `gemini.vertex.batch_prediction.create` (batch.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/docs/predictions/get-batch-predictions)
* `gemini.vertex.datasets` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/vertex-ai/docs/datasets/datasets)
* `gemini.vertex.endpoints.create` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/vertex-ai/docs/predictions/use-private-endpoints)
* `gemini.vertex.experiments.create` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/docs/experiments/intro-vertex-ai-experiments)
* `gemini.vertex.metadata` (agent.adk\_runtime) — [vendor docs](https://cloud.google.com/vertex-ai/docs/ml-metadata/introduction)

## Vertex RAG Engine

**Capabilities** (17/24 rows usable):

* `gemini.vertex.rag_engine` (vector\_store.rag\_engine) — [vendor docs](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/rag-engine/rag-overview)
* `gemini.vertex.rag.augment_prompt` (vector\_store.search) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/rag-engine/retrieval)
* `gemini.vertex.rag.chunking_config` (vector\_store.import) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/rag-engine/parse-and-chunk)
* `gemini.vertex.rag.corpora.create` (vector\_store.create) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora/create)
* `gemini.vertex.rag.corpora.get` (vector\_store.get) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora/get)
* `gemini.vertex.rag.corpora.list` (vector\_store.list) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora/list)
* `gemini.vertex.rag.corpora.patch` (vector\_store.create) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora/patch)
* `gemini.vertex.rag.embedding_model_config` (vector\_store.create) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/rag-engine/manage-your-rag-corpus)
* `gemini.vertex.rag.evaluation` (eval.vertex) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/rag-engine/rag-evaluation)
* `gemini.vertex.rag.files.delete` (vector\_store.documents.delete) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora.ragFiles/delete)
* `gemini.vertex.rag.files.get` (vector\_store.documents.get) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora.ragFiles/get)
* `gemini.vertex.rag.files.import` (vector\_store.import) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora.ragFiles/import)
* `gemini.vertex.rag.files.list` (vector\_store.documents.list) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora.ragFiles/list)
* `gemini.vertex.rag.files.upload` (vector\_store.upload) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/projects.locations.ragCorpora.ragFiles/upload)
* `gemini.vertex.rag.retrieve_contexts` (vector\_store.search) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/rag-engine/retrieval)
* `gemini.vertex.rag.transformations` (vector\_store.import) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/rag-engine/parse-and-chunk)
* `gemini.vertex.rag.vector_db.ragmanaged_db` (vector\_store.create) — [vendor docs](https://cloud.google.com/vertex-ai/generative-ai/docs/rag-engine/manage-your-rag-corpus)

## Vertex Vector Search

**Capabilities** (5/7 rows usable):

* `gemini.vertex.vector_search.endpoints.create` (vector\_store.create) — [vendor docs](https://cloud.google.com/vertex-ai/docs/reference/rest/v1/projects.locations.indexEndpoints/create)
* `gemini.vertex.vector_search.indexes.create` (vector\_store.create) — [vendor docs](https://cloud.google.com/vertex-ai/docs/reference/rest/v1/projects.locations.indexes/create)
* `gemini.vertex.vector_search.indexes.delete` (vector\_store.delete) — [vendor docs](https://cloud.google.com/vertex-ai/docs/reference/rest/v1/projects.locations.indexes/delete)
* `gemini.vertex.vector_search.indexes.get` (vector\_store.get) — [vendor docs](https://cloud.google.com/vertex-ai/docs/reference/rest/v1/projects.locations.indexes/get)
* `gemini.vertex.vector_search.indexes.list` (vector\_store.list) — [vendor docs](https://cloud.google.com/vertex-ai/docs/reference/rest/v1/projects.locations.indexes/list)

## Video generation

**Capabilities** (13/18 rows usable):

* `gemini.veo.aspect_ratio` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.audio_generation` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.duration` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.generate_videos` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.image_to_video` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.params.number_of_videos` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.params.person_generation` (provider.safety\_filters) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.params.resolution` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.params.seed` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.predict_long_running` (media.video.predict\_lro) — [vendor docs](https://ai.google.dev/api/models)
* `gemini.veo.veo_2` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.veo_3` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)
* `gemini.veo.veo_3_fast` (media.video.generate) — [vendor docs](https://ai.google.dev/gemini-api/docs/video)

**Parameters** (13):

| Parameter | Type   | Default | Allowed | Risk | Notes                                               |
| --------- | ------ | ------- | ------- | ---- | --------------------------------------------------- |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/api/models)            |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |
| `option`  | object | —       | —       | low  | [docs](https://ai.google.dev/gemini-api/docs/video) |