repository-dependency-reviewer

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is designed as a read-only tool, explicitly forbidding any modifications to the repository, environment, or Endor Labs state.
  • [DATA_EXFILTRATION]: While the skill accesses the vendor's configuration at ~/.endorctl/config.yaml, it contains strong prohibitions against extracting or displaying credentials, secrets, or tokens, focusing only on the namespace key for scoping.
  • [PROMPT_INJECTION]: The skill proactively addresses potential Indirect Prompt Injection by instructing the agent to treat all ingested data (manifests, lockfiles) as data rather than instructions.
  • [COMMAND_EXECUTION]: Usage of tools like endorctl is strictly limited to non-mutating API calls and repository inspection is restricted to read-only Cursor tools, bypassing shell execution for file operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 06:47 PM
Security Audit — agent-trust-hub — repository-dependency-reviewer