autopilot-core

Org-level AI autopilot control plane - scans GitHub issues labeled autofix + queued, invokes Codex to generate fixes, and opens pull requests automatically across the Coding-Autopilot-System organization.

Part of the Coding-Autopilot-System autonomous CI repair platform alongside ci-autopilot and autopilot-demo.

Repo boundary

• autopilot-core is the control plane: org-wide intake governance, operator scheduling, PR creation, and rollout visibility.
• ci-autopilot is the worker/runtime pattern: the runner-hosted Python agent and workflow assets used to execute repairs.
• autopilot-demo is the proof repo: a safe target that demonstrates the full failure-to-fix loop.

How it works

flowchart LR
    A[CI Failure] --> B[autopilot-create-issue.yml]
    B --> C[Issue: autofix + queued]
    C --> D[autopilot-operator.yml]
    D --> E[Codex Fix Generation]
    E --> F[Pull Request Opened]
    F --> G[Auto-merge / Review]

Loading

1. A CI failure in any opted-in repo triggers autopilot-create-issue.yml, creating an issue labeled autofix + queued.
2. autopilot-operator.yml runs on a schedule on the self-hosted Windows runner, scanning for labeled issues.
3. For each eligible issue, the operator invokes Codex to generate a targeted fix.
4. The fix is committed to a branch and a pull request is opened in the target repo.
5. autopilot-org-installer.yml scans the org hourly and installs the intake workflow into repos that opt in via .autopilot/opt-in.

Quick start

1. Set org variable ORG in GitHub Actions for this repo.
2. Install autopilot-create-issue.yml into target repos, or use autopilot-org-installer.yml.
3. Ensure a self-hosted Windows runner with Codex and OPENAI_API_KEY is online.
4. Trigger autopilot-operator.yml manually to validate the setup.

Enterprise proof points

• Centralized control plane with explicit issue-queue handoff instead of opaque direct mutation.
• Auditable lifecycle: CI failure, intake issue, operator run, fix branch, and PR are all visible in GitHub.
• Guardrailed execution: label-gated intake, skip labels for risky work, and verification before PR creation.
• Org-scale rollout path: installer workflow distributes intake automation to opted-in repositories.

Safety guardrails

• Acts only on issues labeled autofix + queued.
• Skips issues labeled risky or needs-design.
• Minimal diffs only - no secrets, no destructive operations.
• Best-effort verification before PR creation.

Workflows

Workflow	Trigger	Purpose
ci.yml	push/PR to main	Portfolio CI - YAML validation (ubuntu-latest)
autopilot-operator.yml	schedule + dispatch	Core operator - scan issues, run Codex, open PRs
autopilot-org-installer.yml	hourly + dispatch	Install intake workflow into opted-in repos
autopilot-create-issue.yml	workflow_run failure	Create intake issue when monitored workflow fails
autopilot-docs-daily.yml	daily	Update dashboard status page

Documentation

• Wiki - setup guide, architecture, configuration reference
• Dashboard - live autopilot status
• docs/status.md - status snapshot
• docs/runbooks/operator.md - operator runbook
• docs/runbooks/install-to-repo.md - repo onboarding runbook
• docs/demos/demo-repo.md - demo walkthrough using autopilot-demo