Model Context Protocol (MCP) server

What it is

Aclamos ships a Model Context Protocol server at /api/mcp/v1. Connect Claude Desktop, ChatGPT, Cursor, or any MCP-capable client to your org and ask natural-language questions or run actions — "How many nominations does the Best Documentary category have?", "Send the finalist email blast to category cat_abc."

Available on Studio and Enterprise plans. Studio gets 10 read-only tools; Enterprise unlocks 3 mutation tools.

Setup (Claude Desktop)

1. Open Settings → API → Model Context Protocol in your Aclamos dashboard.
2. Click Generate connection token. Copy the resulting bearer token (it's shown once).
3. In Claude Desktop, open Settings → MCP servers → Add server:

- Server URL: https://aclamos.app/api/mcp/v1 - Bearer token: paste the token from step 2

1. Restart Claude Desktop. Ask "what shows do I have?" — Claude should call list_shows and return a table.

Setup (ChatGPT / Cursor / other MCP clients)

Same URL + bearer token. Specifics vary by client; refer to your client's MCP setup docs.

Tools available on Studio

Read-only tools — safe to expose to any AI client:

• list_shows, get_show
• list_nominations, get_nomination
• list_categories, get_category
• list_judges, get_judge
• list_reviews
• get_org_summary

Each tool returns JSON shaped to the same schema as the public REST API; the AI doesn't need a custom integration to consume the results.

Mutation tools (Enterprise only)

Three mutation tools, each requiring confirm: true on every invocation:

• create_judge — create + invite a judge
• assign_judge_to_category — wire a judge to a category for review
• send_email_blast — send a templated email to a list of entrants

Every mutation is recorded in the audit log with the AI client's bearer token id, so you can answer "who/what triggered this?" weeks later. Mutations from MCP clients respect the same RBAC rules as the dashboard — if your token's user doesn't have the capability, the call fails.

Rate limits

120 calls per minute per org. Bursts above that return 429 with a Retry-After header. Studio plans get the same per-minute limit as Enterprise; the difference is purely which tools are registered.

Security

• Tokens are stored as HMAC-SHA256 hashes; the raw token is shown only once at generation.
• Tokens have a configurable expiry (default 90 days); rotate yours from the dashboard.
• Revoke a token instantly from Settings → API → Active tokens.
• Every MCP call is logged with token id, tool name, parameters, and outcome.

Cost

MCP usage is included in your plan. The underlying AI calls (Claude reasoning over your data) come from your AI credit pool — same billing as in-app AI features.