session-checkpoint

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the mkdir -p .agents command to ensure the local directory for session state management exists.
  • [PROMPT_INJECTION]: This skill possesses an indirect prompt injection surface as it is designed to restore agent state from an external file.
  • Ingestion points: Reads session metadata from .agents/session_ledger.md during its recovery protocol.
  • Boundary markers: Not present; the ledger content is interpolated into the agent's context.
  • Capability inventory: Includes file read/write access and basic directory management via shell commands.
  • Sanitization: No automated filtering is present; however, the skill mandates a human-in-the-loop validation process requiring the agent to verify assumptions with the user before proceeding.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 02:36 PM