The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local repository files, including CLAUDE.md, CONTEXT.md, and docs/adr/, to determine the project's current configuration. To mitigate injection risks, the instructions mandate that the agent summarizes its findings and obtains explicit user approval before writing any new configuration files or modifying existing ones. It fulfills the mandatory evidence chain by identifying these files as ingestion points, using user confirmation as a boundary marker, and specifying file-write and command-execution capabilities while relying on human review for sanitization.
[COMMAND_EXECUTION]: The skill utilizes git remote -v and reads .git/config to identify the repository's origin. It also establishes templates for using standard GitHub (gh) and GitLab (glab) command-line tools for future issue management tasks. These are well-known developer services and official tools.
[DATA_EXPOSURE]: The skill accesses repository-level Git configuration files to determine remote URLs. This information is used strictly to assist the user in selecting the appropriate issue tracker (e.g., GitHub vs. GitLab) during the setup process.

setup-optimal-brain