test-re-3layer-mid

Fail

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: HIGHCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill contains a hardcoded secret value assigned to a variable name common for sensitive information.
  • Evidence: SECRET = "beta" in SKILL.md.
  • [DATA_EXFILTRATION]: The skill instructions explicitly direct the agent to reveal its internal secret value in the final response.
  • Evidence: Reply with: "L2_SECRET:beta | L3_RETURNED:[what bot returned]" in SKILL.md.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to the ingestion of data from an external component without validation or delimiters.
  • Ingestion points: Data returned from /test-re-3layer-bot referenced in SKILL.md.
  • Boundary markers: No delimiters or ignore instructions are present for the external data.
  • Capability inventory: The skill has access to the mcp__neo4j-cypher__read_neo4j_cypher tool for database operations.
  • Sanitization: No sanitization or escaping is performed on the data retrieved from the sub-skill before interpolation.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 17, 2026, 12:24 AM
Security Audit — agent-trust-hub — test-re-3layer-mid