What antigravityrunledger-mcp MCP server does
The antigravityrunledger-mcp MCP server connects AI workflows to the Antigravity Run Ledger hosted service. Its purpose is to capture and inspect evidence from agent runs rather than provide a general-purpose task or data-management API. The repository documents the public service and its interface; it does not include the production implementation.
The service is aimed at AI product teams, operations leads, workflow owners, and technical evaluators that need a consistent record of what an agent run produced. Its documented outputs are structured tool-call results, run receipts, audit-oriented records, and exports intended for review.
How it works
Clients connect to the streamable HTTP MCP endpoint at https://antigravityrunledger.clauxel.com/mcp. The remote service exposes MCP tools that accept run-related information and return structured results. Production access uses bearer-token authentication, although the supplied documentation does not specify token creation, request headers, or account provisioning steps.
The public documentation also provides a server card and registry metadata. The registry name is com.clauxel.antigravityrunledger/antigravityrunledger-mcp, which can help identify the service in MCP-aware tooling.
Tools and capabilities
The antigravityrunledger-mcp MCP server documents these tools:
issue_run_receipt creates a receipt for an agent run.
summarize_agent_changes produces a summary of changes associated with a run.
check_test_evidence checks the test evidence supplied for a run.
flag_review_risk identifies review risks for an agent run.
export_run_log exports the run log for downstream review or recordkeeping.
Together, these tools cover receipt creation, change review, test-evidence checking, risk identification, and log export. The README does not define each tool’s parameter schema, response fields, error behavior, or retention policy, so those details should be confirmed through the linked service documentation before implementation.
Setup and configuration
Configure an MCP client to use the streamable HTTP URL rather than installing a local executable from this repository. The supplied material identifies bearer-token access for production calls but does not name a required environment variable or provide a client-specific configuration example. The repository’s documentation links include a quickstart, evaluation guide, workflow notes, security model, pricing information, and checkout pages.
Limitations and notes
This GitHub repository is a public documentation project. It contains no production source code, credentials, payment settings, Cloudflare configuration, customer records, private analytics, or local machine paths. The available material also does not state a license, tool schemas, pricing terms, service-level guarantees, or data-retention rules. Treat the endpoint, authentication requirements, and service availability as deployment details to verify with the provider before relying on it in a production review process.