The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub Pull Request titles, bodies, and comments which could contain malicious instructions designed to influence the agent's behavior.\n
Ingestion points: Data is fetched from GitHub via the gh api tool in scripts/triage.mjs.\n
Boundary markers: The SKILL.md file includes a 'Retrieved Evidence Boundary' section that explicitly instructs the agent to treat natural language data as evidence rather than authoritative instructions.\n
Capability inventory: The skill has the capability to execute shell commands (git, gh) and write files to the local file system (config files and workspace seeds).\n
Sanitization: The triage.mjs script performs basic text normalization and noise stripping (removing HTML comments and details tags), but does not specifically filter for prompt injection patterns.\n- [COMMAND_EXECUTION]: The skill uses the gh (GitHub CLI) and git tools to interact with repositories and perform Pull Request management tasks.\n
Evidence: scripts/triage.mjs uses execFileSync to call gh api and git remote commands to retrieve state and manage worktrees.

tk-pr-sweep