checkpoint

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests information about files modified and session progress, which could theoretically contain malicious instructions. However, the risk is low as it is used for local state persistence within a structured YAML format.
  • Ingestion points: Git diff output and file metadata tracked during the session.
  • Boundary markers: Data is structured into a YAML schema, separating different types of state information.
  • Capability inventory: The skill utilizes Write, Edit, and restricted Bash tools.
  • Sanitization: Instructions require the agent to generate summaries and logic rather than verbatim content insertion.
  • [COMMAND_EXECUTION]: The skill uses specific Bash commands (git, date, wc) to gather session metadata. These are restricted to the allowed-tools list and are used for their intended purposes of identifying changed files and generating timestamps.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 06:16 AM
Security Audit — agent-trust-hub — checkpoint