The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands using execFileSync and spawnSync to perform its core functions. These include running git diff to identify changes, gh pr view to fetch existing pull request descriptions, and gh pr edit to update them with visual recaps. It also invokes a local Node.js script (tools/check-mermaid-syntax.ts) to validate diagram syntax before submission.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data in the form of GitHub Pull Request descriptions, which could contain malicious instructions. While this represents an attack surface, the risk is minimized as the skill primarily performs text substitution between fixed markers.
Ingestion points: upsert-recap-block.mjs retrieves the current PR body via the gh CLI.
Boundary markers: The skill uses  and  tags to isolate its generated content from the rest of the PR description.
Capability inventory: The skill has the capability to read local files, execute git commands, and modify remote PR descriptions using the user's authenticated GitHub session.
Sanitization: No specific sanitization or escaping is applied to the ingested PR body text, but it is handled as a raw string for replacement logic.

visual-recap