The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local commands using pnpm to run build scripts (pnpm -F site api-docs) and start a development server (pnpm dev). It also leverages the GitHub CLI (gh pr list) and git log to extract additional context for documentation. These actions are standard for the skill's intended purpose within a developer workflow.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the ingestion of untrusted data from GitHub Pull Requests.
Ingestion points: The agent is instructed in SKILL.md (Step 1.6) to retrieve context from linked PRs using gh pr list --search.
Boundary markers: Absent; no specific delimiters or instructions are provided to help the agent distinguish between its core instructions and potentially adversarial content within a PR description.
Capability inventory: The agent has permissions to read source files, write MDX files to the project repository, and execute local shell commands (pnpm, gh, git).
Sanitization: Absent; there is no validation or filtering process described for the content fetched from the external GitHub API before it is processed by the agent.

api-reference