The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various gh (GitHub CLI) and git commands to inspect repository metadata, list and create labels, and commit configuration files. This includes using gh repo view to determine the default branch and gh label create to install a workflow taxonomy.
[PROMPT_INJECTION]: The skill implements logic for indirect prompt injection (Category 8) by ingesting data from untrusted repository files to determine its configuration:
Ingestion points: The skill reads package.json, Makefile, Cargo.toml, go.mod, pyproject.toml, and .github/workflows/*.yml (SKILL.md).
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands within the files being read.
Capability inventory: The skill has the capability to write to the file system (creating .ai/agentic.config.json and directories) and execute shell commands via gh and git (SKILL.md).
Sanitization: While the skill typically confirms detected settings with the user, the --defaults flag bypasses this validation, allowing potentially malicious strings detected in repository files to be written directly into the pipeline configuration, which other skills in the collection then use to execute shell commands.

setup-agent-pipeline