The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill configures project-level tools using standard package managers (npx, uvx, npm). These include the context7 MCP server from Upstash (a well-known technology service) and other development-oriented utilities like code-review-graph and Agentation, which are associated with the vendor's ecosystem.
[COMMAND_EXECUTION]: The orchestrator automates repository management tasks using standard Git commands (init, clone, remote, commit) and shell scripts. It also manages project-level configuration updates and installs local git hooks for documentation staging.
[DATA_EXFILTRATION]: The skill promotes secure data handling by automatically generating .env.example files and ensuring sensitive patterns (secrets, environment variables) are added to the .gitignore during project initialization to prevent accidental exposure.
[PROMPT_INJECTION]: The skill processes untrusted external data (project source code, database schemas, and API routes) to maintain its documentation sync.
Ingestion points: Reads code from project repositories during initialization (init-flow.md) and when refreshing documentation (refresh-flow.md, doc-templates.md).
Boundary markers: Absent when scanning source code for entity extraction.
Capability inventory: The skill possesses powerful capabilities including Bash, Write, Edit, and the ability to spawn sub-agents (SKILL.md).
Sanitization: No explicit sanitization or filtering of codebase content is mentioned before it is processed by the agent.

project