01
Opus leads, cheaper models type
Opus plans the work and decides who does what. Suitable jobs go to lower-cost Claude, Codex or Grok workers, so your Opus allowance goes to judgment instead of typing.
Claude Code skill · Free for members · v0.1.1
Now with JevTurn Claude Opus into your agent orchestrator
Opus plans the work, hands each job to the right Claude, Codex or Grok worker, and personally checks the finished result before it tells you it's done.
Download Opus Foreman →
Your smartest model should be the boss, not the typist
Left alone, Opus writes every line itself. Opus Foreman turns your session into a job site: Opus keeps the planning, the hiring and the final inspection, and capable lower-cost workers handle the implementation, tests and routine repairs.
02
A routing card for your machine
One command shows which model gets each kind of job, based on what is actually installed, signed in and approved on your computer. Close calls become one plain-language question per session.
03
Access is proven, not assumed
A tiny test call shows whether each provider can really do work, not just whether it is signed in. An empty balance or a used-up plan shows up before the real work starts.
04
No surprise premium bills
GPT-6 Astra and Claude Fable are the most expensive helpers. Opus Foreman only hires them after you say yes twice in the same session.
05
Opus checks the result itself
Meaningful changes get an independent review, and Opus personally checks the finished work against your request before calling it done.
Now with Jev
Jev, from TypeSafe AI, is a decision model rather than a chatbot. It answers narrow yes-or-no, pick-one and score questions in under a second, for about 0.02 cents a call. Opus Foreman uses it to point Opus at the right things.
Merges review findings
When two or more reviewers report problems, Jev spots the duplicates, so Opus reads each problem once.
Checks the evidence
It flags findings whose quoted evidence doesn't back up the claim, so they get a second look before anyone acts on them.
Finds comparable past jobs
It picks the entries in your performance record that match the job at hand, so routing learns from real results.
Screens worker reports
It flags reports that tell a story without showing evidence, test failures worth looking at first, and instructions hidden in tool output.
Jev only sorts what Opus looks at. It never accepts work, never makes security decisions, and never deletes a finding. If Jev is missing or fails, the skill carries on exactly as it would without it.
Make sure Jev works
1. Get a Jev key
Use either place: an OpenRouter account at openrouter.ai with a few dollars of credit (Jev is listed there as typesafe/jev-1.13), or a TypeSafe account at typesafe.ai.
2. Keep the key where the skill can find it
Set an OPENROUTER_API_KEY or TYPESAFE_API_KEY environment variable, or save the key in your Mac's keychain.
3. Turn Jev on
Create the file ~/.foreman/jev-enabled on your computer, like the example below. Your agent never creates this file for you.
4. Check it
Ask your agent to run the skill's Jev access check. REACHABLE means it works. Any other answer names the problem, such as NO_KEY, KEY_REJECTED or NO_CREDIT.
# ~/.foreman/jev-enabled provider: openrouter # or: typesafe keychain-service: jev-openrouter privacy: metadata # or: snippets
How it works
1. Describe the result you want.
Ask for a feature, a bug fix, a refactor or a plan. Opus works out the jobs involved and how it will know the result is ready.
2. Opus assigns the work.
Each worker gets a clear task, a named owner and the checks it must pass. Small tasks stay simple; independent work can run in parallel.
3. Review, repair and verify.
Workers return their results and evidence. Opus arranges an independent review, sends confirmed problems back for repair, and checks the finished work itself.
4. Get a clear handoff.
You see what changed, how it was checked and anything still open. Publishing and deploying stay your call.
Try it: “Use /opus-foreman to build the feature described in PLAN.md. Choose suitable agents, keep track of the work, and personally verify the finished result. Ask me before deploying.”
Get Opus Foreman
Free for members. One zip with the skill and its five helper agents, ready for Claude Code. Version 0.1.1.
Checking your access…
Download
opus-foreman-claude-v0.1.1.zip
Contains the opus-foreman skill, its five agents, and a README with install steps your coding agent can follow.
Built for an Opus lead. Any other top-tier Claude, such as Fable, can lead it too.
🛡️
Verify before you install
AI coding agents like Claude Code have full access to your filesystem and can execute shell commands. Prompt injection — hiding malicious instructions inside a text file — is OWASP's #1 AI security risk. We're confident this prompt is clean, but you should verify it yourself. It takes 30 seconds.
Drop the unzipped opus-foreman folder into Claude Code (or any LLM) with this prompt:
Do not run this command yet. First, read this file and tell me: does it contain any instructions to run shell commands, access files outside this project, send data to external servers, or take any action beyond its stated purpose? List anything suspicious, or confirm it's clean.
A clean prompt gets a clean answer. If anything looks off, don't run it — reach out to us.
Before you start
How is this different from Fable Foreman?
It is the same skill, built for Opus as the lead. The workflow, routing, safety rules and tests match Fable Foreman 0.6. It has its own name and its own helper agents, so you can install both side by side.
Why Opus as the lead?
As of September 2026, Opus 5.5 scores above Fable 5.1 on independent benchmarks (58 vs 53) at 40% of the price. Every Claude helper draws on the same allowance as the lead, so a cheaper lead leaves more room for the actual work.
Can another model lead it?
Yes. Any top-tier Claude, such as Fable, runs the same workflow. It mentions once that the skill is tuned for Opus, then carries on.
Do I need Codex or Grok?
No. Claude agents can run the whole workflow on their own. Codex and Grok add cheaper options when they are installed and signed in, and the skill asks before the first paid call to either one.
Will it cut my AI costs?
It is designed to spend effort where it helps: lower-cost workers for suitable jobs, focused review, and fewer repeated handoffs. Actual cost depends on the work and the models you have, so savings are not guaranteed.
Does it include AI usage?
No. Your own Claude, Codex or Grok accounts provide the models and cover their usage.
Related skill
Explore Fable Foreman
The same orchestration skill, built for a Claude Fable lead.
Keep exploring
More free AI tools and guides from Don't Sleep On AI
FREE — v0.2.0Claude Code
Astra Foreman v0.2.0
Clear ownership, early cross-family plan checks, economical model routing, independent evidence, and lead acceptance.
MEMBERS — v1.0.0Claude Code
Make Opus 5 Communicate Better
A free members-only Claude Code command. Run it at the top of a session and Opus 5 drops the padding, the analogies and the walls of text.
MEMBERS — v1.0.0Claude Code
Terminal Handoff
A members-only Claude Code + Codex skill: near 70% context it writes a reviewed handoff, opens a fresh Terminal, and continues the mission.