Alfred turns Claude Code and Codex into a spec-driven engineering team. Named agents plan the work, write the code, review each other, and open pull requests. On your machine, behind an approval gate.
Alfred Desktop is the recommended onboarding path. It detects an existing install, starts or connects to the local server, checks GitHub and engine auth, guides repo selection, and lets you choose a roster theme or custom display names. Slack is still where teams discuss and approve plans.
About a minute of Alfred working on a repo. It answers a question about the code, turns a change you describe into a plan you approve, and scopes the same request from Slack. Its agents then build it, review it, and open a pull request.
Each agent has one plain job, and the default install wires the whole engineering loop: planning, architecture, implementation, review, test coverage, high-priority review fixes, cleanup, summaries, and safe automerge when your policy allows it.
The public OSS architect role. Reads the parent issue or Slack plan, checks the affected repos, gets approval, and files scoped repo-level work for the implementation fleet.
[10:30] architect drafted bundle:billing-v2 [10:31] architect posted plan to Slack [10:34] scope approved [10:35] architect opened backend -> web -> mobile tasks
Picks up a scoped task, makes its own copy of the repo, runs Claude Code or Codex with the full project in context, and opens a pull request with tests.
[12:04] senior-dev claimed your-backend#247 [12:04] senior-dev worktree ~/.alfred/wt/12042118/ [12:18] senior-dev PR #1247 opened
Reads a Slack request, GitHub issue, or plain-text plan, asks for missing details, then files small tasks a coding agent can pick up on its own.
[09:02] drake refined settings plan [09:02] drake filed your-backend#248 agent:implement [09:03] drake linked acceptance checks
Reads the changes, runs your test suite in its own copy of the repo, leaves comments on the lines that need work, and marks the pull request ready or needs changes.
[12:18] ras-al reviewed PR#1247 [12:18] ras-al filed 2 inline nits [12:19] ras-al labelled needs-changes
Looks at recently changed files, finds the places that need coverage, and opens focused test PRs that stand on their own.
[15:10] bane found low coverage in billing [15:23] bane PR #1250 opened [15:24] bane labelled tests-only
Finds agent-authored PRs with unresolved high-priority review comments, applies the fix in a clean worktree, and pushes the update back to the PR branch.
[16:01] nightwing found P1 on PR#1247 [16:08] nightwing pushed fix [16:09] nightwing asked for re-review
Alfred checks the Slack drafts, plans, and GitHub issues you marked for it, works in isolated repo copies, opens pull requests, routes review, and comes back to Slack when it needs a decision.
Alfred finds the next task, hands it to the right role, runs the work in an isolated repo copy, routes review, records memory, and comes back to Slack when a human decision is needed.
A Mac mini, an old MacBook on a dock, or a small Linux box is enough. Alfred runs locally, using the Claude Code and Codex subscriptions you already pay for.
Planning happens in a Slack thread. Reply with what you want, what should change, or questions of your own, and Alfred turns the discussion into the next task. Nothing big runs until you approve it.
Alfred keeps notes about each repo on your machine. Structural gates and an LLM judge decide what becomes durable context, and you can still inspect or override the memory queue from Slack or the desktop app.
Alfred can plan a single feature across several repos at the same time. It works the same way inside a monorepo, whether you use Turborepo, Nx, or Cargo workspaces.
Every line is on GitHub at github.com/luminik-io/alfred. Fork it, read it, run it locally, or audit the code.
Alfred keeps notes on your own machine: durable lessons, file touches, failure patterns, what went wrong before, and what is in progress right now. Memory promotion is designed to run without a human babysitting every candidate: structural gates and an LLM judge decide what becomes durable context, while you can still inspect or override it.
Agents and trusted teammates can suggest a lesson worth keeping. Alfred filters candidates through structural checks and an LLM judge before durable promotion, so memory keeps improving without turning every operator into a queue janitor.
Setup blockers, timeouts, and repeated test failures all land in one place. The next time something breaks, your agents start from what happened before instead of digging back through old Slack messages.
The default recall path is a local Redis Agent Memory Server, with FleetBrain as the reliability ledger and inspection surface. If Redis is disabled, missing, or unreachable later, Alfred keeps running and exposes Redis-specific checks through the memory controls instead of treating FleetBrain as the same thing.
The same suite of ten repo-convention tasks plus two controls ran twice
through the claude CLI, changing only whether memory was on. It
measures one thing: whether an agent repeats a mistake the fleet already
learned about.
Method and repro.
Start with a Slack thread, a GitHub issue, or a plain-text plan. Batman is the default-theme name for the public architect role: it handles approved multi-repo fan-out, Lucius handles repo-local implementation as senior-dev, and Drake turns smaller requests into clear tasks.
Batman is the default-theme name for the architect role, and that role is public OSS code. It reads the parent issue, checks the affected repos, plans the rollout, gets approval in Slack or the local client, and files scoped child issues. Senior-dev, test-engineer, fixer, reviewers, and the merge gate then finish each PR.
From the moment a run starts to the pull request it opens. The same five stages happen every time an agent fires.
Slack is the planning and control surface. Trusted users can start requests, refine plans in a thread, approve big work, ask for status, pause agents, and keep context attached to the work. Slack never bypasses the approval gates.
Real output from a local OSS full-fleet install on 2026-06-30: architect, senior-dev, planner, test-engineer, fixer, triage, ops-watch, e2e-runner, memory, code-map, cleanup, reports, and fleet doctor are loaded from the start.
Alfred
alfred status output from a real Mac, captured 2026-06-30 after the OSS full-fleet install. Cost is $0.00 because Alfred runs on your own Claude and Codex subscriptions, not pay-per-token API bills. Batman, Huntress, Gordon, Bane, Nightwing, and the support jobs are loaded from the first install.
Run doctor for setup health, then dry-run one agent to see the path it would take without touching GitHub, Slack, an engine, or your files.
In the last 30 days, Alfred helped merge 10 PRs across 9 Luminik repos. The impact page shows aggregate product totals and public examples.