The Agent Skills Directory

[COMMAND_EXECUTION]: Python scripts including create_handoff.py, check_staleness.py, and setup_test_env.py utilize the subprocess module to execute git commands. These calls gather project history, branch details, and file status to populate handoff documents. The implementation uses argument lists rather than shell strings.
[INDIRECT_PROMPT_INJECTION]: The skill reads handoff documents that contain externally-sourced data from the project's Git history and file structure.
Ingestion points: The RESUME Workflow in SKILL.md directs the agent to read handoff files from the .claude/handoffs/ folder.
Boundary markers: Metadata is inserted into the handoff template without explicit markers to distinguish between data and instructions.
Capability inventory: The skill provides functionality for file system access and Git command execution via subprocess.
Sanitization: The validate_handoff.py script performs scans for secrets and TODO placeholders but does not filter natural language content for potential injection attacks.

session-handoff