The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/ingest_source.py executes a local validation script scripts/validate_governance.py using subprocess.run. This is a controlled execution of a component bundled with the skill to verify staged changes before committing them.
[INDIRECT_PROMPT_INJECTION]: The skill processes external documents from the sources/ directory. While this presents an attack surface where malicious content in a source could attempt to influence the agent's concept creation, the risk is mitigated by mandatory operator approval of takeaways and structural validation of the generated proposal.
Ingestion points: Files located in the sources/ and wiki/ directories are read by the agent.
Boundary markers: Use of structured JSON proposals and YAML frontmatter templates to delimit generated content.
Capability inventory: File system write access within the wiki root and execution of the local validator script.
Sanitization: The require_confined function enforces that all file operations are restricted to the designated wiki directory, effectively preventing directory traversal attacks.

okf-wiki-ingest