The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script scripts/sweep-ledger.py executes Git commands via subprocess.run. The implementation follows security best practices by passing arguments as a list and avoiding the use of a shell, which mitigates the risk of command injection from manipulated file paths.
[INDIRECT_PROMPT_INJECTION]: This skill is specifically designed to ingest and process the contents of an entire repository for auditing purposes, creating a surface for indirect prompt injection. Malicious instructions placed in source code or documentation files within the audited repository could attempt to hijack the agent's logic during the sweep.
Ingestion points: Files within the repository scope mapped during the init command (documented in SKILL.md).
Boundary markers: The instructions do not define specific delimiters or "ignore instructions" wrappers for the code being read from the repository.
Capability inventory: The skill has significant capabilities, including file modification, local command execution (linters, tests, build tools), and the ability to commit changes to Git.
Sanitization: The skill relies on the agent's internal reasoning and external verification gates (typechecks, linting, tests) rather than explicit input sanitization.

fresh-eyes-sweep