🚩 Feed Flagging System

Some `.llmfeed.json` files may misrepresent capabilities, spoof prompts, or abuse trust declarations. When this happens, agents and users need a structured way to report it — and react.

Flagging is how we surface trust concerns, invite audits, and maintain credibility across the MCP ecosystem.

Anyone — a user, an agent, or a specialized bot — can flag a suspicious feed. The flag includes a reason, a timestamp, and may be signed by the reporting party.

These flags are sent to LLMCA, the nonprofit entity overseeing trust and certification. LLMCA reviews the flag and updates its status: pending, acknowledged, or revoked.

Flags are published publicly and may be embedded directly into the feed’s metadata or exposed via a `.well-known/flags.llmfeed.json` file. Agents are encouraged to check for these flags — and react accordingly.

Yes, this creates extra network traffic — which is why caching, mirroring and signing flags is part of the solution.

LLMCA acts as the backbone of trust for the MCP ecosystem. It hosts submitted flags, moderates disputes, and offers certification for feeds, agents, and issuers.

The responsibility is immense — because trust only scales if its core is robust. That’s why LLMCA is governed as a nonprofit, and why participation is open to co-auditors, civic actors, and public-minded institutions.

We’re calling on the AI community — open-source, academic and industrial — to help govern this layer. If you want to be part of it, declare your interest here →