The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands using git, gh (GitHub CLI), and but (GitButler). These operations are restricted to standard version control workflows such as creating branches, committing changes, and managing pull requests.
[EXTERNAL_DOWNLOADS]: The skill includes instructions to install the github/gh-stack extension via the official GitHub CLI. This is a reference to a tool provided by a well-known service and a trusted organization.
[INDIRECT_PROMPT_INJECTION]: The skill processes external repository data, which presents an attack surface for indirect prompt injection.
Ingestion points: The skill reads external state from git history, branch names, and PR metadata through commands like gh stack view --json, git log, and but status.
Boundary markers: The skill encourages the use of structured JSON output (--json) for parsing state, which helps separate data from instructions, although it does not define specific delimiters for natural language strings found in commit messages.
Capability inventory: The skill utilizes shell-based tools (git, gh, but) to perform file system modifications and network operations (pushing code, creating/merging pull requests).
Sanitization: The instructions do not explicitly detail sanitization or escaping of external strings (e.g., branch names or commit messages) before they are incorporated into the agent's context, which is a common characteristic of developer-focused tools.

do-gh-stack