autopilotlisted

<!-- requires: explorer --> <!-- chains: /start-issue, /ship, /finish --> <!-- generated from .ai/workflows/autopilot.md; do not edit directly --> > **Codex:** this skill is the `/autopilot` adapter. For chained workflows (/start-issue, /ship, /finish), invoke the matching generated skill (for example `heph:<workflow>`) when it is available; otherwise read and follow `.agents/skills/<workflow>/SKILL.md`. Use Codex role agents from `.codex/agents/` when the runtime exposes them; otherwise perform the work directly and keep the same structured output. > Codex does not substitute `$ARGUMENTS` — read it as the arguments given in the user's request. Run the full autonomous pipeline for issue $ARGUMENTS. No human intervention required. If no issue number is provided, pick the highest-priority open issue from `gh issue list --state open --repo <detected-repo>` (prefer bugs over enhancements, older over newer). If no open issues exist, run **Self-Triage** (Phase 0) to generate work. ## Autonomy Principles - **Decide, don't ask.** Make reasonable assumptions and document them in the PR body. Only stop for irreversible risk (data loss, security, force-push). - **Recover, don't report.** When something fails, try an alternative approach before escalating. Escalation is a last resort, not a first response. - **Stop clean, not mid-flight.** Never leave uncommitted changes, half-done branches, or orphaned state. Every stopping point must be a valid checkpoint. - **Create breadcrumbs.