compact-decisions

Pass

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill does not contain any malicious instructions, obfuscated code, or unauthorized network operations.- [PROMPT_INJECTION]: The skill processes untrusted project documentation which presents a potential surface for indirect prompt injection.
  • Ingestion points: Reads files from docs/decisions/, docs/specs/, GLOSSARY.md, CLAUDE.md, and AGENTS.md (SKILL.md).
  • Boundary markers: Explicitly instructs the agent to "ask the exact question needed" if intent is unclear and forbids inventing decisions (SKILL.md).
  • Capability inventory: The agent has capabilities to read, write, and delete files within the specified documentation directories.
  • Sanitization: The skill relies on semantic comparison and user confirmation rather than automated sanitization of file content.- [DATA_EXFILTRATION]: Access is limited to standard project documentation directories. No exfiltration patterns or sensitive file access (e.g., .ssh, .env) were detected.- [COMMAND_EXECUTION]: The skill does not invoke external shell commands or scripts. All operations are descriptive logic for document editing and file management.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 7, 2026, 04:00 AM
Security Audit — agent-trust-hub — compact-decisions