The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of shell commands through the Bash tool and generated Makefiles. It executes commands for repository structure auditing (git log), site building (pnpm/astro), and deployment (rsync).
[EXTERNAL_DOWNLOADS]: During the scaffolding and publishing phases, the skill instructs the agent to use pnpm/npm to install dependencies. These include well-known packages such as @astrojs/starlight and Astro from the official npm registry.
[DATA_EXFILTRATION]: The skill generates deployment configurations (Makefile) that utilize rsync to transfer documentation artifacts from the local site/dist/ directory to a user-defined remote VPS host. This is a standard operation for documentation publishing.
[PROMPT_INJECTION]: The /docs review-drafts subcommand implements an indirect prompt injection surface. It employs sub-agents to read untrusted project data and verify documentation accuracy.
Ingestion points: Project source code located in backend/, app/, and src/ directories, as well as CHANGELOG.md files.
Boundary markers: None identified; sub-agents are directly exposed to raw file content.
Capability inventory: Sub-agents are explicitly granted the capability to modify documentation files using the Edit tool based on their analysis of the ingested code.
Sanitization: No sanitization or validation of the repository content is performed before it is processed by the sub-agents.

docs