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

# MCP Calls

> Audit MCP tool invocations across the workspace — which tool, which agent, what arguments, what outcome.

# MCP Calls

`cmd/crewship/cmd_admin_extras.go` wraps `GET /api/v1/mcp-tool-calls`. Lists recent MCP tool invocations across the workspace — useful for tracking which external integrations are doing real work and which are dormant.

## `crewship mcp-calls`

Each row in the response carries the tool name, the agent that called it, the arguments, and the outcome.

| Flag              | Type   | Default | Effect                                                                                                                                             |
| ----------------- | ------ | ------- | -------------------------------------------------------------------------------------------------------------------------------------------------- |
| `--limit <n>`     | int    | `50`    | Max calls to return. Sent as `?limit=`.                                                                                                            |
| `--filter <expr>` | string | (unset) | Pipe the JSON output through the `jq` binary (must be installed and on `PATH`). If `jq` is missing, the raw, unfiltered output is printed instead. |

### Examples

```bash theme={null}
crewship mcp-calls
crewship mcp-calls --limit 200
crewship mcp-calls --filter '.[] | select(.tool == "linear:create_issue")'
crewship mcp-calls --format yaml
```

Default output is indented JSON — the rows don't have a canonical table form, so JSON + `jq` is the most useful pretty default.

## See also

* [`crewship mcp`](/cli/mcp) — install and configure MCP integrations.
* [`crewship integration`](/cli/integration) — broader integration management.
* [`crewship audit`](/cli/audit) — workspace audit trail (MCP calls are also journaled).