The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements an Indirect Prompt Injection surface by allowing AI agents to process untrusted external data from GitHub Issues and PR comments. Ingestion points: The 1-implement.yml and 2-fix-review.yml workflows ingest GitHub Issue descriptions and PR review comments as instructions for the AI agent. Boundary markers: No explicit delimiters or warnings are provided to the agent to differentiate between the system's pipeline instructions and the content of the issues or comments. Capability inventory: The orchestrated agent uses the claude-code-action with tools including Bash, Read, Write, Edit, Glob, and Grep. It also requires GitHub Action permissions to create and push commits to repository branches. Sanitization: No sanitization or filtering is performed on the incoming issue or comment text before it is presented to the agent.
[COMMAND_EXECUTION]: The skill requires and orchestrates the use of system-level CLI tools and GitHub Action configurations with high privileges. The pipeline relies on the git and gh (GitHub CLI) utilities for repository management and workflow dispatching. The setup instructions require enabling high-privilege repository settings, such as allowing GitHub Actions to create and approve pull requests and enabling auto-merge.

agent-pipeline