The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes content from Markdown and MDX files within the documentation directory to perform its validation gates. This creates an attack surface where malicious instructions embedded in the documentation files could influence the agent's behavior during the audit, reporting, or fixing phases.
Ingestion points: Markdown files located in docs/src/content/docs/ (SKILL.md).
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands were identified.
Capability inventory: The skill has the ability to execute local Python scripts, shell commands via Bun, and perform file writes.
Sanitization: No sanitization or validation of the ingested Markdown content was identified.
[COMMAND_EXECUTION]: The skill executes several command-line tools and local scripts provided within the repository.
Python Scripts: scripts/progress_read.py, scripts/validate_glossary.py, and scripts/term_read.py are executed using the uv tool.
Build Tools: The skill runs bun install, bun run build, and bun run verify-search within the docs directory to generate the site and validate the search index.
[EXTERNAL_DOWNLOADS]: The execution of bun install facilitates the download and installation of external dependencies from the NPM registry, which is a well-known package service.

final-proofread