refactor-project
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes the entire project codebase, which could contain malicious instructions hidden in comments or strings. * Ingestion points: All code files discovered in SKILL.md Phase 1 and processed by the refactor:code-simplifier subagent in Phase 2. * Boundary markers: Absent. The prompt context defined in references/agent-configuration.md does not instruct the subagent to treat file contents as data rather than instructions or to ignore embedded commands. * Capability inventory: The skill and its subagent utilize Task, Read, Edit, Glob, Grep, and Bash(git:*) tools. * Sanitization: Absent. There is no evidence of filtering or escaping for content read from the codebase.
- [COMMAND_EXECUTION]: The skill recommends the use of git reset --hard HEAD as a rollback mechanism in references/output-requirements.md and references/agent-configuration.md. This is a destructive command that will discard all uncommitted changes in the working directory without further warning.
Audit Metadata