The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external and untrusted sources such as Git commits, pull requests, and project documentation. This data is used to inform the agent's summary and trigger maintenance actions, creating a risk that malicious content in a PR or commit could influence the agent.
Ingestion points: masterplan.md, changelog, Git status, commits, pull requests, and AGENTS.md.
Boundary markers: Absent; the skill does not specify delimiters or instructions to ignore embedded commands in the files it reads.
Capability inventory: Shell script execution (sh) and Python script execution (python3).
Sanitization: Absent; the skill does not mention validating or escaping the content read from files or Git logs before processing.
[COMMAND_EXECUTION]: The skill requires the execution of several local scripts to refresh plans and check for kit updates.
Evidence: Instructions to run sh .agents/tools/plan-refresh.sh, python3 maintain/scripts/upgrade-check.py, and sh records/scripts/trued-mark.sh.

what-now