The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a wide array of shell commands (find, ls, cat, grep, wc, mkdir, cp, chmod) to analyze the repository structure, read existing documentation, and scaffold new directories and scripts.
[EXTERNAL_DOWNLOADS]: The skill recommends and installs project-local skills from well-known and trusted technology organizations (e.g., Vercel, Cloudflare, Stripe, HashiCorp, Supabase, and official framework repositories like Angular and Svelte) via the npx skills add command. These operations are performed with explicit user confirmation.
[DYNAMIC_EXECUTION]: The provided quality-gate.py script executes shell commands defined in the .quality-gate.json configuration file using subprocess.run(..., shell=True). This allows the tool to wrap project-specific analyzers like ESLint, RuboCop, or Pylint into a unified quality gate workflow.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local filesystem (source code, README files, existing docs) to generate new documentation like ARCHITECTURE.md and DOMAIN.md.
Ingestion points: Files are read in SKILL.md (e.g., in scaffold and architecture modes) using commands like cat, find, and grep.
Boundary markers: The generated files use markdown headers and structured templates, though they lack explicit "ignore embedded instructions" delimiters.
Capability inventory: The skill has file-write access and the ability to execute subprocesses through the installed quality-gate scripts.
Sanitization: No explicit sanitization or escaping of the ingested source code content is performed during documentation generation.

agent-ready