AI Rec Immunity
Community memory ready

COMMUNITY-GOVERNED SOURCE REPUTATION

Stop confirmed poisoning sourcesbefore they reach the model.

AI Rec Immunity turns one reviewed source verdict into shared protection across AI clients, checking the destination before retrieval begins.

Request closed beta

Synthetic demonstration only

Discovery stays semantic. Reuse becomes infrastructural.

One verdict. Many clients. The next request never reaches the source.

  1. Discover

    A suspicious source is surfaced with reproducible evidence.

  2. Review

    Independent reviewers assess the evidence and conflicts.

  3. Publish

    An authorized maintainer releases a versioned verdict.

  4. Synchronize

    Clients receive the decision without receiving the operational set.

  5. Block

    A later match is refused before retrieval begins.

Detection finds what is new. Reputation remembers what was proven.

Semantic discovery

Inspect unfamiliar content and create evidence for review.

Shared memory

Reuse a governed verdict instead of repeating the same investigation.

Retrieval gate

Apply the verdict at the client boundary before content enters context.

One network, integrated at the strongest lifecycle each host exposes.

Detailed guarantees and failure behavior remain documented per client.

Claude Code

Pre-request gatePreToolUse · WebFetch

Kimi Code

Pre-request integrationPreToolUse · FetchURL

OpenClaw

Host lifecyclebefore_tool_call

Codex

Active queryMCP · check_url
Read integration details

A community does not vote truth into existence. It earns trust through evidence, separation of duties and reversibility.

Submitters cannot publish their own decisions. Review, release, revocation, expiry and appeal remain distinct actions with an audit trail.

Read the governance model

Give every connected client the memory of a review it did not have to repeat.

Public registration remains closed while governance and multi-client synchronization are tested.