The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of "project's declared cleanup commands" during the post-merge cleanup phase. These commands are typically sourced from the repository's own configuration, creating a vector for executing arbitrary local commands based on the project environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from repository files and commit history to synchronize with issue trackers, which serves as an injection surface.
Ingestion points: The agent is instructed to read contents from docs/specs/<slug>/spec.md and Spec-Folder trailers within commit messages.
Boundary markers: No specific delimiters or boundary markers are mentioned to differentiate instructions from data in these files.
Capability inventory: The skill utilizes gh pr merge for GitHub operations, performs local branch/worktree deletion, and executes project-specific cleanup scripts.
Sanitization: There is no evidence of sanitization or validation performed on the issue IDs, spec folder paths, or folder content extracted from the branch.
[REMOTE_CODE_EXECUTION]: The skill instructs the agent to provide a command for installing missing dependencies: npx skills@latest add toy-crane/skills --skill <name>. This involves downloading and executing the skills package from the NPM registry at runtime.
[EXTERNAL_DOWNLOADS]: The skill relies on external dependencies (pull, pr, clean-branches) and provides instructions to download them via the NPM registry if they are missing from the environment.

merge