skills/jcosta33/skills/bulletproof/Gen Agent Trust Hub

bulletproof

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and verify external 'claims' and 'targets', creating a vulnerability surface where malicious instructions could be embedded in the data being verified.\n
  • Ingestion points: The 'Method' section in SKILL.md describes freezing and processing a 'claim set' and 'target' provided in the context.\n
  • Boundary markers: The skill includes logic-based boundaries by instructing the agent to 'Define proof and falsification before searching' and 'Freeze the verification target'.\n
  • Capability inventory: The skill leverages sensitive capabilities such as local file system inspection ('Inspect authoritative local sources') and shell command execution ('Run safe checks', 'Run the matching branch') as defined in SKILL.md.\n
  • Sanitization: There is no evidence of specific input sanitization, filtering, or escaping for the external claims processed by the methodology.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 09:14 PM
Security Audit — agent-trust-hub — bulletproof