## Background Model output is untrusted but may be treated as safe. ## Why it matters Prompt injection can mislead tooling. ## Acceptance criteria - [ ] Document that model output is untrusted; never execute it. - [ ] Add to SECURITY.md. ## Where to look - `SECURITY.md` ## How to claim Comment `/claim` on this issue.
Background
Model output is untrusted but may be treated as safe.
Why it matters
Prompt injection can mislead tooling.
Acceptance criteria
Where to look
SECURITY.mdHow to claim
Comment
/claimon this issue.