plan-fix

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh issue view) to retrieve issue metadata and standard Git commands (git branch, git commit) for managing the local development environment. These operations are restricted to local execution and do not involve pushing to remote repositories.
  • [PROMPT_INJECTION]: The skill processes untrusted external data from GitHub issues, which presents a surface for indirect prompt injection.
  • Ingestion points: External data is ingested from GitHub issue titles, bodies, and comments via the gh issue view command as specified in references/PLANNING_PROCESS.md.
  • Boundary markers: The agent is instructed to follow a 'Senior software architect' persona and uses strict output templates (Hand-off blocks) to maintain consistent behavior.
  • Capability inventory: The skill is capable of writing documentation files (SPEC.md, ACCEPTANCE.md) and performing local Git operations (branching and committing).
  • Sanitization: The skill enforces numeric validation for issue tokens and transforms issue titles into kebab-case topics to sanitize inputs used in branch naming and file paths.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 11:08 PM
Security Audit — agent-trust-hub — plan-fix