The Agent Skills Directory

[COMMAND_EXECUTION]: The file anti_slop_brain/cli.py contains calls to subprocess.call() to execute internal scripts from the scripts/ directory. These calls are defensive, using the current Python executable (sys.executable) and passing arguments as a list rather than a single string, which prevents shell injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection. (1) Ingestion point: untrusted data enters context via scripts/ingest_source.py (triggered by the ingest command). (2) Boundary markers: absent in the provided CLI code; metadata instructions rely on YAML structure. (3) Capability inventory: internal script execution via subprocess.call in anti_slop_brain/cli.py. (4) Sanitization: absent in the provided CLI wrapper. The risk is mitigated by strong behavioral guardrails in agents/anti-slop-secretary.md and wiki/meta/CONVENTIONS.md that mandate citation and grounding in verified vault notes.
[SAFE]: The project demonstrates excellent security hygiene, including a comprehensive security policy (SECURITY.md), publishing guidelines (PUBLISHING_NOTICE.md) that explicitly forbid the storage of credentials, and release checks (RELEASE_CHECKLIST.md) that verify the absence of secrets and local absolute paths.
[SAFE]: According to the documentation and third-party notices, the skill relies solely on the Python standard library for its runtime operations, which significantly reduces supply chain risks.

anti-slop-brain