skills/metroscope-dev/skills/clean-up/Gen Agent Trust Hub

clean-up

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to process external, potentially untrusted data from the repository to identify artifacts for deletion, creating an indirect prompt injection surface.
  • Ingestion points: Repository implementation diffs, status reports, ADRs, CONTEXT.md files, and various temporary documentation files (SKILL.md).
  • Boundary markers: None identified; instructions do not specify delimiters or warnings to ignore instructions embedded in the processed files.
  • Capability inventory: The agent is authorized to perform file deletion ("Delete all repository-local artifacts"), file modification ("Update a relevant README.md"), and shell command execution ("run the repository's documentation checks, formatters, or tests") (SKILL.md).
  • Sanitization: None identified; content from files is processed directly without escaping or validation.
  • [COMMAND_EXECUTION]: The skill directs the agent to execute local shell commands, including git diff --check and arbitrary repository-specific formatters or tests, which could be exploited if those local tools are compromised (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 04:24 PM
Security Audit — agent-trust-hub — clean-up