What it does
This MCP server lets AI agents audit their own EU AI Act compliance and issue signed certificates. It provides HMAC-signed audit logs, hash-chained records that the publisher describes as tamper-proof, A2A agent tracking and compliance reporting. The publisher states that it supports Articles 9, 13, 15, 26 and 50 of the Act, where applicable.
It suits developers building agents that need a verifiable record of their compliance checks, or teams that want audit trails covering agent-to-agent activity. The tools named in the publisher's examples are self_audit, audit_conversation and get_certificate.
It can be added to the MCP configuration file for Claude Desktop, or to another MCP client's configuration. Alternatively, it can be installed with pip and started as a stdio-transport command. The free self-hosted version is available, and the publisher says that paid tiers add attestations that can be verified through an external service.
Limitations and requirements
Verifiable attestations are described as part of paid tiers above the free self-hosted version. Article coverage is stated as applicable only where relevant. The server requires an MCP client, such as Claude Desktop, to be configured and running.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
io.github.CSOAI-ORG/ai-self-audit-mcp
Ai Self Audit MCP Server by MEOK AI Labs
v1.0.14As shown on registry.modelcontextprotocol.io when we last read it