The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources to define its orchestration logic and branch strategy.
Ingestion points: The references/github-autopilot.md script fetches issue body and title content via gh issue view.
Boundary markers: There are no explicit instructions to the agent to ignore embedded commands or malicious directives within the parsed issue content.
Capability inventory: The skill possesses high-privilege permissions to write to worktrees, manage branches, and modify pull requests and issue trackers.
Sanitization: It uses regex to extract branch names from the body, but the broader orchestration wave logic is driven by external prose.
[COMMAND_EXECUTION]: The skill relies on complex shell-based logic to perform repository and state management.
Evidence: references/github-autopilot.md contains a multi-step bash script that executes gh, git, jq, awk, and sed using data derived from external issue specifications.

orchestrate