The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches skill definitions, frontmatter, and repository content from agentskill.sh and GitHub (raw.githubusercontent.com) to support the research and exploration workflows.
[COMMAND_EXECUTION]: Executes npx commands to interact with the 'skills' CLI and the vendor-scoped '@codevoyant/agent-kit' for tasks including plan registration, notification, and project scaffolding.
[PROMPT_INJECTION]: The skill exhibits a surface area for indirect prompt injection as it retrieves and processes content from untrusted external repositories and user-supplied URLs.
Ingestion points: workflows/explore.md (Agent B/C fetching GitHub/agentskill.sh content) and workflows/new.md/workflows/update.md (processing user-provided research URLs via the WebFetch tool).
Boundary markers: The researcher agent prompt (agents/skill-researcher.md) explicitly instructs the model to be 'completely faithful to the source' and 'not invent, infer, or explore beyond what is written'.
Capability inventory: The skill is capable of writing files to the local system, executing shell commands through npx, and spawning multiple subagents with specific model assignments (Opus/Sonnet/Haiku).
Sanitization: The workflow requires multiple AskUserQuestion confirmation steps, ensuring the user reviews and accepts the generated design plans before any permanent file changes or tool installations are performed.

skill