the-warden

Pass

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill performs static analysis on local project data. There is no evidence of credential harvesting, obfuscation, or external network requests beyond expected lookups for dependency deprecation information.
  • [INDIRECT_PROMPT_INJECTION]: The skill maintains a potential surface for indirect prompt injection because it processes untrusted source code and dependency manifests.
  • Ingestion points: Reads project source files, dependency manifests (package.json, requirements.txt, go.mod, Cargo.toml), and internal documentation (SKILL.md).
  • Boundary markers: Absent; the skill does not instruct the agent to use delimiters or ignore instructions found within analyzed code blocks.
  • Capability inventory: Limited to file system read access and git diff execution; no network exfiltration or file system write instructions are present.
  • Sanitization: No sanitization is performed on the ingested code or metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 8, 2026, 10:05 AM
Security Audit — agent-trust-hub — the-warden