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

# Tool Profiles

> Reduce schema overhead with role-based tool subsets.

Tool profiles control which tools are visible to a connected agent. By limiting the tool set to what's relevant, you reduce schema overhead and keep each session focused on only the capabilities it needs.

## Questions This Page Answers

* Which profile should I use for my operator workflow?
* How much tool-surface reduction does each profile provide?
* Why do narrower profiles improve reliability and token efficiency?

## Why Profiles Matter

* **Lower schema load**: fewer tools means clearer selection and lower token overhead.
* **Safer execution**: profiles reduce accidental high-impact calls in the wrong context.
* **Faster operator loops**: less noise in suggestions, cleaner control-plane behavior.

## How to Use Profiles

Append `?profile=NAME` to the MCP connection URL:

```
https://mcp.useorgx.com/mcp?profile=commander
```

Or in your MCP config:

```json theme={"dark"}
{
  "mcpServers": {
    "orgx": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.useorgx.com/mcp?profile=executor"]
    }
  }
}
```

Default is `full` — all tools registered, backward compatible.

***

## Memory Profile

**Purpose**: Shared organizational memory for Claude, ChatGPT, Cursor, and
other MCP clients.

**Tools**:

| Category  | Tools                                                                              |
| --------- | ---------------------------------------------------------------------------------- |
| Bootstrap | `orgx_bootstrap`, `orgx_describe_tool`, `orgx_describe_action`, `workspace`        |
| Memory    | `orgx_decide`, `orgx_search`, `orgx_search`, `get_decision_history`, `orgx_decide` |
| Approvals | `orgx_decide`, `orgx_decide`, `orgx_decide`, `orgx_decide`                         |
| Context   | `orgx_inspect`, `orgx_recommend`, `orgx_recommend`, `orgx_recommend`               |

**Best for**: Directory users and first-time installs that want OrgX to behave
like shared company memory before exposing the full planning and execution
surface.

<Tip>
  Start with `memory` when your first prompts are "remember this decision",
  "what did we decide", "find that artifact", or "show work waiting on me".
  Switch to `full` when you want planning, task delegation, widgets, and
  initiative execution.
</Tip>

***

## Commander Profile

**Purpose**: Human operators managing decisions, entities, and org overview.

**Tools** (27):

| Category       | Tools                                                                                                                    |
| -------------- | ------------------------------------------------------------------------------------------------------------------------ |
| Decisions      | `orgx_search`, `orgx_decide`, `orgx_decide`                                                                              |
| Agents         | `orgx_spawn`                                                                                                             |
| Memory         | `orgx_search`                                                                                                            |
| Monitoring     | `orgx_recommend`, `orgx_recommend`, `orgx_recommend`                                                                     |
| Entity CRUD    | `orgx_search`, `orgx_write`, `update_entity`, `orgx_act`, `orgx_act dry_run=true`, `scaffold_initiative`, `orgx_inspect` |
| Comments       | `comment_on_entity`, `list_entity_comments`                                                                              |
| Prioritization | `orgx_recommend`, `record_quality_score`                                                                                 |
| Workspace      | `workspace`                                                                                                              |
| Onboarding     | `configure_org`                                                                                                          |
| Stats          | `stats`                                                                                                                  |

**Best for**: ChatGPT users, dashboard operators, team leads reviewing agent output.

<Info>
  The commander profile still carries legacy compatibility aliases such as
  `orgx_decide`, `get_decision_history`, and `orgx_recommend` for
  older clients. New operator flows should prefer `orgx_search`,
  `orgx_search`, and `orgx_recommend`.
</Info>

***

## Planner Profile

**Purpose**: Initiative planning, hierarchy scaffolding, and plan sessions.

**Tools** (18):

| Category        | Tools                                                                         |
| --------------- | ----------------------------------------------------------------------------- |
| Planning        | `orgx_plan`, `get_active_sessions`, `orgx_plan`, `orgx_plan`, `orgx_plan`     |
| Entity Creation | `scaffold_initiative`, `orgx_write`, `update_entity`, `batch_create_entities` |
| Browse          | `orgx_search`, `orgx_inspect`, `orgx_act`                                     |
| Context         | `orgx_search`, `orgx_recommend`, `orgx_recommend`                             |
| Workspace       | `workspace`                                                                   |

**Best for**: Product managers defining initiatives, planning agents, strategy sessions.

***

## Executor Profile

**Purpose**: Agent execution — progress reporting, changesets, spawning.

**Tools** (14):

| Category      | Tools                                                             |
| ------------- | ----------------------------------------------------------------- |
| Telemetry     | `orgx_emit_activity`, `orgx_apply_changeset`, `sync_client_state` |
| Authorization | `orgx_spawn`, `orgx_spawn`                                        |
| Streams       | `update_stream_progress`, `get_initiative_stream_state`           |
| Tasks         | `orgx_act`, `orgx_search`, `orgx_inspect`, `comment_on_entity`    |
| Agents        | `orgx_spawn`, `orgx_spawn`                                        |
| Workspace     | `workspace`                                                       |

**Best for**: Background agent workers, CI/CD pipelines, automated task execution.

***

## Observer Profile

**Purpose**: Read-only monitoring and reporting.

**Tools** (18):

| Category       | Tools                                                         |
| -------------- | ------------------------------------------------------------- |
| Decisions      | `orgx_search`                                                 |
| Agents         | `orgx_spawn`                                                  |
| Monitoring     | `orgx_recommend`, `orgx_recommend`, `orgx_recommend`          |
| Memory         | `orgx_search`                                                 |
| Browse         | `orgx_search`, `orgx_inspect`, `list_entity_comments`         |
| Prioritization | `orgx_recommend`, `get_scoring_signals`, `get_scoring_config` |
| Streams        | `get_initiative_stream_state`                                 |
| Workspace      | `workspace`                                                   |
| Stats          | `stats`                                                       |

**Best for**: Dashboards, reporting tools, status monitors, read-only integrations.

<Info>
  The observer profile also keeps `orgx_decide`,
  `get_decision_history`, and `orgx_recommend` registered as compatibility
  aliases. Use `orgx_search`, `orgx_search`, and `orgx_recommend`
  in new recipes.
</Info>

***

## Full Profile (Default)

All tools registered. No restrictions. Use when you need maximum flexibility or are exploring the API.

**Token impact**: \~2x the tokens of a focused profile.

***

## Profile Size Reduction

Compared to the full tool surface:

| Profile   | Tools   | Approx. Reduction |
| --------- | ------- | ----------------- |
| Memory    | Focused | Highest           |
| Commander | 27      | \~56%             |
| Planner   | 18      | \~70%             |
| Executor  | 14      | \~77%             |
| Observer  | 18      | \~70%             |
| Full      | All     | 0% (baseline)     |

<Tip>
  Start with the most restrictive profile that covers your use case. You can
  always switch to `full` if you need additional tools.
</Tip>

***

## Choosing a Profile

| Use Case                             | Recommended Profile |
| ------------------------------------ | ------------------- |
| Shared company memory for agents     | `memory`            |
| Human reviewing decisions in ChatGPT | `commander`         |
| Building initiative plans in Cursor  | `planner`           |
| Background agent executing tasks     | `executor`          |
| Dashboard or monitoring integration  | `observer`          |
| Exploring the API / development      | `full`              |

***

## Next Steps

<CardGroup cols={2}>
  <Card title="Agent Quickstart" icon="rocket" href="/docs/agent-ops/agent-quickstart">
    Get connected with your chosen profile.
  </Card>

  <Card title="Full Tool Reference" icon="wrench" href="/docs/api/mcp-tools">
    See all tools across all profiles.
  </Card>
</CardGroup>
