Alfred
VerifiedAutonomous agents that plan, code, and review using your existing subscriptions.
What is Alfred?
Alfred is an open-source system that turns your paid Claude Code and Codex access into a background engineering team. It monitors repositories, assigns narrow roles to different agents, and moves work through standard GitHub issues and pull requests while sending updates to Slack.
The workflow starts with planning agents creating scoped tasks, followed by implementation in fresh worktrees, automated reviews, and test additions. All activity stays local, respects spend limits, and requires explicit approval gates before autonomous execution begins.
It suits engineering teams that want recurring maintenance and feature work handled without constant oversight, especially those already subscribed to frontier coding models and comfortable running local automation.
What you can build with Alfred
Multi-repo rollouts
Batman coordinates changes across several repositories while keeping each agent scoped to a single worktree and reporting status back to a shared Slack thread.
Ongoing code maintenance
Drake generates implementation issues from specs, Lucius writes the code, Ra's al Ghul reviews it, and Bane adds tests without requiring the operator to stay at the keyboard.
Slack-driven planning
Team members reply in plan threads to adjust scope or add criteria, with changes persisted locally and surfaced for operator approval before any code is written.
Install Alfred
git clone https://github.com/luminik-io/alfred-os.gitgit clone https://github.com/luminik-io/alfred-os.git ~/code/alfred-os
cd ~/code/alfred-os
PYTHONPATH=lib python3 examples/bin/echo_summarise.py --dry-run- 1Clone the repository to your local machine.
- 2Navigate into the project directory.
- 3Run the dry-run command to observe the full lifecycle with no external calls.
- 4Configure GitHub, Slack, and local CLI auth for your chosen models.
- 5Start the scheduler so agents can pick up new issues automatically.
Alfred: pros & cons
Pros
- +Uses subscriptions you already pay for instead of new API billing
- +Keeps all execution local with explicit spend and approval limits
- +Coordinates through familiar GitHub and Slack primitives
- +Assigns narrow, non-overlapping roles to each agent
Cons
- –Requires a machine that stays online to run the fleet
- –Setup involves multiple local CLIs and permissions
- –Still early-stage with limited public usage examples
Frequently asked questions
No, it shells out to your local Claude and Codex CLI tools so you stay within the subscriptions you already have.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar agents
Other general-purpose options worth comparing.
