runtime-auditor

Pass

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
  • [NO_CODE]: The skill consists entirely of instructional documentation and metadata. No executable code, shell scripts, or tool definitions are provided.- [PROMPT_INJECTION]: The skill instructions (in SKILL.md and meta/skill.json) direct the agent to process untrusted data sources, including "logs" and "user reports." This ingestion point creates a surface for indirect prompt injection if the audited data contains hidden commands.
  • Ingestion points: "logs", "traces", "alerts", and "user reports" mentioned in SKILL.md.
  • Boundary markers: No delimiters or "ignore instructions" warnings are suggested for processing external data.
  • Capability inventory: The skill is instructional only and does not provide new capabilities, though it guides the use of the agent's existing capabilities.
  • Sanitization: No sanitization or data validation steps are included in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 31, 2026, 11:46 AM
Security Audit — agent-trust-hub — runtime-auditor