loop-constraints

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and obey rules from a local file (loop-constraints.md), which is a common pattern for defining agent boundaries.
  • Ingestion points: Reads loop-constraints.md from the project root.
  • Boundary markers: None; instructions in the file are treated as binding constraints for the agent's session.
  • Capability inventory: The skill influences triage priority and limits file modification capabilities of other skills (e.g., minimal-fix).
  • Sanitization: No specific sanitization or escaping is mentioned for the ingested rules.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill identifies and protects sensitive data by establishing a default denylist for critical project paths when no custom configuration is found.
  • Evidence: The skill explicitly prohibits editing of .env, .env.*, auth/, payments/, secrets/, and credentials/ directories, reducing the risk of accidental secret exposure.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 01:02 PM
Security Audit — agent-trust-hub — loop-constraints