Documentation

Install the GitHub App, add .dz/maintainer/maintainer.yml, and watch the first loop run.

The product, screen by screen, in the order work moves: Overview, Plans, Fleet, CI runs, Pending actions, Blockers and Audit log.

Every policy field, with examples. The file that bounds what the agent may do.

How a repository is recognised, the one-command trio it earns, and the gate exit codes that mean "could not grade".

All MCP tools (user-tier and operator-tier) with scopes and actions.

Mint a personal access token, point Claude Code, Cursor, or tesote.ai at mcp.developerz.ai, and dispatch your first task.

Hand fixes to Copilot via webhook. The agent opens a PR; your policy decides what happens next.

Trigger Claude Code on triaged issues. Full audit trail, every tool call logged.

Route tasks to OpenHands for autonomous coding. BYOK: your key, your bill.

CodeRabbit-compatible AI review on your own model. Drop in your .coderabbit.yaml or use the native review: block.

Move review onto your own key without rewriting config: your .coderabbit.yaml keeps working.

Dependabot and Renovate pull requests reviewed, fixed and merged through the merge gate, plus a recommended dependabot.yml.

Remind, then close, issues and pull requests that wait on the person who opened them. Off until you set it, with the settings we recommend.

Replace GitHub-hosted runners with your own VPS. One label change, runs-on: [self-hosted, linux, developerz].

Enrol a VM into your fleet: requirements, the join command, hardening, and what never touches the box.

Agents store files in your S3 bucket or the platform bucket. Presigned URLs, expiry by policy.

Raw CI/review/run logs are kept 72 hours. Point a Postgres you own at the platform for retention on your terms.

JSON-defined boards pushed live over SSE. Define from Claude Code or Claude Desktop.

Programmatic access to your repos, tasks, runners, billing, and audit log.

Signed event deliveries to your endpoint: envelope, signature, retries, and which events fire today.

Sentry issue alerts become asks that triage can turn into fix pull requests, merged on your policy.

Per-token read and write budgets, the pre-auth per-IP bucket, and exactly what a 429 looks like.

When the agent pages a human: security, breaking changes, and stuck threads.

Re-decide a recorded session against a candidate maintainer.yml. Executes nothing, stores nothing.

Bring your own model key. We orchestrate; you own the inference bill.

Seats, entitlements, spend caps and the Stripe portal, and why inference never appears on our invoice.

Owner, admin, member and viewer: what each can do, how membership is granted, and the last-owner rule.

Least-privilege GitHub scopes, encrypted secrets, and the append-only audit log.

The agent is quiet, the PR will not merge, the box is offline. The checks that resolve most reports.

AI dev, box, seat, lane, handoff, disposition. The words this documentation uses, defined once.