The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to read external files such as STATUS.md, AGENTS.md, README, or handoff documents to determine the next tasks or locate installation commands.\n
Ingestion points: Contents of AGENTS.md, README, STATUS.md, and other local repository documentation are read into the agent's context in Step 5.\n
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present when processing these files.\n
Capability inventory: The agent is authorized to execute shell commands, including git operations and repository-specific installation tools like npm or pnpm.\n
Sanitization: The skill does not define any sanitization or validation logic for the commands or instructions retrieved from the repository files.\n- [DYNAMIC_EXECUTION]: The skill dynamically determines and executes installation commands by parsing repository documentation or identifying tools from lockfiles.\n
Evidence: Step 5 ("Dependencies") instructs the agent to "run the repo's install command" and to "Take the command from its AGENTS.md, README or task runner," which results in the execution of arbitrary commands defined within the repository being cleaned up.\n- [COMMAND_EXECUTION]: The skill uses shell commands and a bundled bash script to perform repository maintenance.\n
Evidence: The agent executes a series of commands including gh pr view, git worktree remove, git branch -D, and git switch, guided by the results of the scripts/classify.sh script.

merge