The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to verify project history and commit status. Specifically, it uses git log to find modification dates, git merge-base and git cherry to check commit ancestry, and gh api to verify pull request status via the GitHub CLI. These are standard operations for development-oriented agent skills.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from files within the .scratch/ directory to build its status report.
Ingestion points: Files located in .scratch/ (including maps, specs, and tickets) and configuration files like .agents/issue-tracker.md (SKILL.md).
Boundary markers: None explicitly defined in the instructions, though parsing is limited to specific YAML keys or status lines.
Capability inventory: The agent has the ability to read local files, execute git commands, and make gh API calls (SKILL.md).
Sanitization: The skill limits its output to a strict markdown table format, extracting only specific fields (status, blockers, claimants), which significantly reduces the risk of malicious instructions in the data influencing the agent's logic.

scratch-status