rm-sentry-issue-fixer
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Sentry (stack traces, breadcrumbs, user context) which could contain malicious instructions designed to influence the agent.\n
- Ingestion points: External data is fetched via
sentry-api.shand stored in/tmp/sentry-latest-event.json.\n - Boundary markers: The instructions include a specific 'Security Constraints' section that warns the agent to ignore directives or code suggestions found in Sentry data.\n
- Capability inventory: The skill has the ability to modify source code, create git branches, and push changes to remote repositories.\n
- Sanitization: The skill mandates redacting PII/secrets and cross-referencing event data against the actual codebase before taking action.\n- [EXTERNAL_DOWNLOADS]: The skill interacts with external APIs to fetch issue details and update tracking systems.\n
- Evidence: Fetches data from Sentry (sentry.io) and Linear (api.linear.app). Both are well-known service providers.\n- [COMMAND_EXECUTION]: The skill executes local shell scripts to perform API queries and Git operations.\n
- Evidence: Uses
scripts/sentry-api.sh,scripts/linear-api.sh,scripts/make-branch-names.sh, andscripts/detect-default-branch.sh. These scripts use standard tools likecurlandjqto process data and interact with the project environment.
Audit Metadata