The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill resolves GitHub issue references using the gh CLI and uses the fetched content as task instructions, creating an attack surface where malicious external data could influence agent behavior.
Ingestion points: SKILL.md (Workflow Step 0) fetches issue data (title, body, labels) via gh issue view.
Boundary markers: Absent. The instructions do not specify delimiters or warnings to ignore instructions within the fetched issue content.
Capability inventory: The skill possesses file editing (Edit), exploration (Glob, Grep), and command execution (lint/typecheck) capabilities.
Sanitization: Absent. The skill uses the raw issue body directly as a task description or implementation checklist.
[COMMAND_EXECUTION]: The skill identifies and executes shell commands discovered within project configuration files to validate code changes.
Evidence: SKILL.md Step 3 (Test) instructs the agent to discover and run lint and typecheck commands from manifest files such as package.json, pyproject.toml, Cargo.toml, or go.mod.

oneshot