Repository navigation
Conversation
Adds a hunter class adapted from jdpru/cover-your-ais (ai-codebase-safety-audit): recursive deletes on a variable, test setup that wipes a database, infrastructure one command can destroy, agents in CI, uncapped money and message sends, and prose-only rules in AGENTS.md, CLAUDE.md and .cursorrules. Test setup files are now read for this class only; prose-rule patterns match instruction files only. Co-authored-by: Atris <299057014+atris-builder[bot]@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Sentinel's audit gets a new thing to hunt: places where one command from a coding agent could destroy data, infrastructure, money or messages with nothing outside the model to stop it.
Adapted from the in-repo checks of jdpru/cover-your-ais, prompts/ai-codebase-safety-audit.md:
The machine-level checks from that list (account tokens, synced folders, shell aliases, MCP pinning, hooks) are not source. Hooks, MCP pinning, permissions and hidden characters are already covered by
atris sentinel agents.Proof: new map test; all 249 sentinel and audit tests pass. A free plan of atrisos-backend seeds 133 blast-radius areas out of 3,921.
🤖 Generated with Claude Code