The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes pull request comment bodies which are supplied by external users, creating a surface for indirect prompt injection.
Ingestion points: In workflows/pr-comments.md, the skill uses the GitHub GraphQL API to fetch the body field of all comments in a pull request review thread.
Boundary markers: None. The instructions do not define delimiters or warnings for the agent to treat the retrieved comment text as untrusted data.
Capability inventory: The skill possesses the capability to execute shell commands via the gh CLI (for API interactions) and git (for code inspection and modification).
Sanitization: No sanitization, escaping, or filtering of the external comment content is performed before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on shell command execution to perform its tasks, using tools like gh, jq, git, and sed.
The workflow in workflows/pr-comments.md constructs shell commands using variables derived from GitHub API responses, such as repository names, PR numbers, and comment IDs.
It uses sed to read local files and git to inspect the codebase, which are standard operations for this skill's stated purpose.

using-github