The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes standard version control and repository management tools (git, gh, glab) to retrieve session context. These commands are used in a read-only manner to inspect repository status, commit logs, and issue tracker lists.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external or shared sources such as Git commit messages, issue descriptions, and codebase comments to populate the handoff document. This represents a potential indirect prompt injection surface if an attacker provides malicious input via these sources that is then summarized into the handoff file for a future agent to read.
Ingestion points: Git commit logs, gh/glab issue tracker output, MEMORY.md, CLAUDE.md, and in-code TODO comments (SKILL.md).
Boundary markers: The skill does not explicitly define boundary markers or sanitization requirements for the content it aggregates from these sources.
Capability inventory: The skill utilizes file system writes (mkdir, file creation) and subprocess execution (git, gh, glab).
Sanitization: The skill relies on the agent to "read the room" and "verify load-bearing claims" but does not include automated sanitization of the ingested content.

handoff