The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill contains hardcoded credentials (admin_user@example.com / TestPassword123!) intended for verifying documentation flows against a local development stack (http://localhost:3000). While specifically for a local environment, hardcoding passwords in instructions is a security risk.
[COMMAND_EXECUTION]: The instructions guide the agent to execute several scripts local to the repository:
scripts/generate.sh for building SVGs and API specs.
python3 scripts/format_docs.py for MDX formatting and link checking.
python3 audit/build_manifest.py for site-wide documentation auditing.
[INDIRECT_PROMPT_INJECTION]: The skill operates on documentation files (MDX) and source code which are external ingestion points.
Ingestion points: Files in the documentation/ directory and various source code files throughout the repository (e.g., web/src/, backend/onyx/).
Boundary markers: None identified in the provided instructions to delimit external content from instructions.
Capability inventory: Execution of shell scripts, python scripts, and pre-commit hooks as described in the Command Execution section.
Sanitization: No explicit sanitization or validation of the ingested documentation content is mentioned before it is processed or audited by the agent.

docs-writing