Show HN: Weir - Catch credential leaks in the AI agents you build and use
Hacker News · 28 Oct 2026, 02:07 AEST
Weir is a tool designed to catch credential leaks in AI agents. It can plant fake credentials to monitor and report any leaks without exposing real values.
Key facts
- Weir is for one problem: credentials crossing the boundaries of your AI agent.
- For agents you build, Weir plants fake credentials where real ones would live, runs the agent's normal flows, and checks what comes out: telemetry, logs, console output, model requests, and requests to other services.
- Right now, this is supported for Claude Code.
- The test mode is deterministic and offline.
- It found GHSA-j7qq-vr8c-xvw3 in MLflow (fixed in 3.17.0).
Read aloud
Weir is a tool designed to catch credential leaks in A I agents. It can plant fake credentials to monitor and report any leaks without exposing real values. Weir is for one problem: credentials crossing the boundaries of your A I agent. For agents you build, Weir plants fake credentials where real ones would live, runs the agent's normal flows, and checks what comes out: telemetry, logs, console output, model requests, and requests to other services. Right now, this is supported for Claude Code. The test mode is deterministic and offline. It found GHSA-j7qq-vr8c-xvw3 in MLflow fixed in three point one seven.zero.
Sources in this cluster
- Show HN: Weir - Catch credential leaks in the AI agents you build and use
Hacker News · 28 Oct 2026, 02:07 AEST