The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes existing repository configuration files which could contain instructions intended to influence the agent's behavior.
Ingestion points: The skill reads .gitattributes, .editorconfig, .gitignore, and .gitmodules from the repository root (SKILL.md, Step 1).
Boundary markers: No explicit delimiters or boundary markers are defined for the content read from these files.
Capability inventory: The skill has the capability to modify repository files (git add --renormalize .), execute GitHub API calls (gh api), and run local Git commands (git lfs, git check-attr).
Sanitization: No specific sanitization or filtering of the ingested file content is mentioned beyond merging it with official templates.
[COMMAND_EXECUTION]: The skill uses the local git and gh CLIs to manage repository state. It includes a specific protocol for line-ending renormalization (git add --renormalize .), which is a destructive operation. Security is managed via internal guardrails: it requires a clean working tree (git status --porcelain) and explicit user authorization before execution.
[EXTERNAL_DOWNLOADS]: The skill fetches official .gitignore templates from GitHub's public API. This is a standard integration with a well-known service to provide valid configuration templates for the user's project.

repo-attributes