# adrata_ai_tool_execute

Execute the same governed AI CRM ToolDispatcher path used by Adrata chat. Free reads run directly. Reads that can spend external data credits return a spend hold; execute one only with confirmSpend:true, a reason, and an idempotencyKey (reuse the same key on retry). The response includes the approved ceiling and an exact provider-reported receipt where available. Writes use a two-step, server-bound confirmationToken flow; a client approved boolean can never authorize a write. A non-interactive machine/OAuth principal (every MCP client) can preview writes but cannot complete one.

Agent View of the PolicyLayer registry record for `adrata_ai_tool_execute`. HTML page: https://policylayer.com/tools/adrata-starfield-mcp/adrata-ai-tool-execute

## Facts

- Tool: `adrata_ai_tool_execute`
- Server: Starfield (`@adrata/starfield-mcp`) — https://policylayer.com/tools/adrata-starfield-mcp.md
- Install: `npx -y @adrata/starfield-mcp`
- Homepage: https://www.npmjs.com/package/@adrata/starfield-mcp
- Risk category: Execute (High risk)
- Registry record: grade F, identity unverified
- Server rate-limited: no
- Parameters: 8 (1 required)
- Recommended policy verdict: Rate-limited

## Parameters

| Parameter | Type | Required | Description |
| --- | --- | --- | --- |
| `dryRun` | boolean | no | Defaults to true for write tools. Set false only after explicit approval. |
| `reason` | string | no | Required audit reason for a spend acknowledgment or when requesting a live-write confirmation token. |
| `approved` | boolean | no | Deprecated compatibility field; never authorizes a live write. |
| `toolName` | string | yes | AI CRM tool name from adrata_ai_tool_catalog, e.g. list_channels or rank_companies_by_icp |
| `arguments` | object | no |  |
| `confirmSpend` | boolean | no | Required true to execute a read held for external vendor spend. Never authorizes a workspace write. |
| `idempotencyKey` | string | no | Required with confirmSpend:true. Reuse the same key on retry; the API replays the first response instead of calling the paid provider twice. |
| `confirmationToken` | string | no | Single-use server token returned by the first live-write request; bound to the user, workspace, tool, and stored arguments. |

Parameters from the server's own tool schema.

## Example call (MCP tools/call, JSON-RPC 2.0)

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "adrata_ai_tool_execute",
    "arguments": {
      "toolName": "<toolName>"
    }
  }
}
```

## Why adrata_ai_tool_execute is rated High

adrata_ai_tool_execute triggers real processes with real consequences. An agent gone sideways doesn't fire it once. It starts dozens of builds, sends mass notifications, or burns through compute before anyone looks up.

Risk signals: Bulk/mass operation — affects multiple targets

## Use case

AI agents invoke adrata_ai_tool_execute to trigger actions in Starfield. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call: builds kicked off, notifications sent, workflows started.

## Recommended policy (PolicyLayer)

Verdict: **Rate-limited**. Enforced by the PolicyLayer MCP gateway (https://policylayer.com/mcp-gateway) before a call reaches Starfield:

```json
{
  "version": "1",
  "default": "deny",
  "tools": {
    "adrata_ai_tool_execute": {
      "limits": [
        {
          "counter": "adrata_ai_tool_execute_rate",
          "window": "minute",
          "max": 10,
          "scope": "grant"
        }
      ]
    }
  }
}
```

## Other tools on Starfield (232)

- `bulk_delete_buyer_groups` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/bulk-delete-buyer-groups.md
- `cancel_agent_task` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/cancel-agent-task.md
- `delete_action` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/delete-action.md
- `delete_buyer_group` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/delete-buyer-group.md
- `delete_company` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/delete-company.md
- `delete_note` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/delete-note.md
- `delete_opportunity` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/delete-opportunity.md
- `delete_partner` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/delete-partner.md
- `delete_person` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/delete-person.md
- `delete_webhook` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/delete-webhook.md
- `flag_work_item` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/flag-work-item.md
- `forget` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/forget.md
- `paper_delete_document` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/paper-delete-document.md
- `paper_revoke_share` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/paper-revoke-share.md
- `remove_buyer_group_member` — Destructive — https://policylayer.com/tools/adrata-starfield-mcp/remove-buyer-group-member.md
- `adrata_api_request` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/adrata-api-request.md
- `attribute_partner_to_deal` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/attribute-partner-to-deal.md
- `build_pursuit_command_center` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/build-pursuit-command-center.md
- `draft_workflow` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/draft-workflow.md
- `dry_run_workflow` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/dry-run-workflow.md
- `record_partner_consumption` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/record-partner-consumption.md
- `replay_workflow_run` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/replay-workflow-run.md
- `request_deployment` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/request-deployment.md
- `request_provider_action_execution` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/request-provider-action-execution.md
- `request_workflow_deployment` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/request-workflow-deployment.md
- `warmup_email` — Execute — https://policylayer.com/tools/adrata-starfield-mcp/warmup-email.md
- `adrata_ai_tool_catalog` — Read — https://policylayer.com/tools/adrata-starfield-mcp/adrata-ai-tool-catalog.md
- `adrata_api_catalog` — Read — https://policylayer.com/tools/adrata-starfield-mcp/adrata-api-catalog.md
- `adrata_desktop_app_audit` — Read — https://policylayer.com/tools/adrata-starfield-mcp/adrata-desktop-app-audit.md
- `check_batch_import_status` — Read — https://policylayer.com/tools/adrata-starfield-mcp/check-batch-import-status.md
- …and 202 more: https://policylayer.com/tools/adrata-starfield-mcp.md

## For agents

This record is a snapshot. Live verdicts and the full registry:

- Check every server in your MCP config at once: `npx -y policylayer stack`
- Vet a server before you add it: install the mcp-precheck skill — `npx skills add https://policylayer.com` (skill text: https://policylayer.com/skill.md)
- Query the registry over MCP: endpoint `https://api.policylayer.com/mcp` — tools `check_mcp_server`, `check_mcp_stack`, `check_tool`, `search_registry`, `get_change_events`

---

Source: the PolicyLayer MCP registry — one continuously verified record per MCP server. Full record: https://policylayer.com/registry?q=adrata-starfield-mcp · API: https://policylayer.com/registry/api · Policy library: https://policylayer.com/policies/adrata-starfield-mcp