The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the GitHub CLI (gh) to inspect repository state and perform merge actions.
Evidence: Shell commands like gh api, gh pr view, and gh pr merge are used to verify permissions, rulesets, and status checks before landing code. These operations are essential for the skill's purpose of enforcing repository governance.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources such as repository-level configuration files and GitHub API responses to resolve autonomy levels.
Ingestion points: Reads .github/lever.yml and repository rules via gh api in SKILL.md (Step 1 and Step 3).
Boundary markers: The skill contains internal logic to ignore instructions from untrusted surfaces like PR bodies or comments, though it lacks explicit prompt delimiters for the configuration file content itself.
Capability inventory: The skill is capable of merging PRs (gh pr merge) if the resolved autonomy level is high enough (L3+).
Sanitization: The instructions focus on logical parsing of YAML fields rather than sanitization of the content, but the risk is mitigated by the skill's restrictive logic regarding what sources can raise authority.

autonomy-dial