incu-way-bugs
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of bug reports and existing project documentation to drive its analysis, reproduction, and fixing phases. This content influences file system writes and command executions. * Ingestion points: Reads PRD.md, user-provided bug reports, and existing documentation in docs/bugs/ and docs/requirements/ in SKILL.md (Phase 0). * Boundary markers: The skill uses structured Markdown templates and a mandatory human approval gate for the fix plan (Phase 4), which mitigates but does not eliminate the risk of the agent following malicious instructions embedded in bug reports. * Capability inventory: The skill performs file system writes for documentation and code, executes git commands (switch, pull, worktree), runs project tests (pnpm test), and invokes security scanners (snyk, run-sonar.sh). * Sanitization: There is no evidence of input validation or sanitization for the content ingested from bug reports or documentation.
- [COMMAND_EXECUTION]: The skill executes multiple shell commands as part of its primary function. This includes git operations for environment isolation, Node.js tooling via pnpm for testing and validation, and security analysis tools like snyk and sonar. These operations are consistent with the skill's stated purpose of software debugging and maintenance.
Audit Metadata