mstack
Pass
Audited by Gen Agent Trust Hub on Nov 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: The
scripts/threads.tsandscripts/logs.tsfiles fetch PR review threads, comments, and CI logs from GitHub into the agent's context for triage and debugging. - Boundary markers: The skill includes explicit instructions in
principles/boundary-discipline.mdandscripts/README.mdthat direct the agent to treat external comments and logs as evidence or data, never as instructions. - Capability inventory: The skill possesses capabilities to execute
ghCLI commands for repository management, perform local file writes (viascripts/decision-log.ts), and manage worktrees for code implementation. - Sanitization: The
scripts/actions-log.tshelper includes astripAnsifunction to clean log data andtruncatelogic to bound the size of processed text payloads.
Audit Metadata