Codex Mind

One guided run, idea to verdict.

Codex Mind chains the two halves of the plugin: it forges a proposal, then puts it through the council — in one guided pass. If you don't have a proposal yet, it asks what to build first. You get one combined cost estimate up front and a single build / revise / stop call at the end.

Every Mind run opens with this banner — the signal that the guided pipeline has begun.

The pipeline

Forge invents. Council judges. Mind runs both.

The Codex Mind pipeline IDEA FORGE COUNCIL BUILD · REVISE · STOP

How the guided run works

Four moves, fully guided.

01

Start

The brain banner appears. If there's no proposal yet, Mind asks one focused question: what should we build, and the hard constraints.

02

Estimate once

One combined token estimate for both stages, shown and accepted up front — so you see the full cost before anything runs.

03

Forge

Five creators shape a bounded proposal (auto second round only on strong discord, hard cap three). You see it before it moves on.

04

Judge & decide

The council reviews the proposal and returns one call — build, revise, or stop — with blockers, dissent, and the exact verification.

Checkpoints, not surprises If Forge can't converge, Mind surfaces the dissent and asks before paying for a council review. If the council finds a blocker, the run stops at revise or stop — never "build" over a live blocker.

When to reach for which

Three doors into the same building.

Mind

You have an idea and want to go all the way to a decision in one guided pass.

Use it

Hand Mind an idea.

In chat — give it something to build, and it runs the whole pipeline:

Prompt Use Codex Mind to forge a proposal for <your idea> and then run it through the council. Show me the combined estimate first, and end with a clear build / revise / stop call.

Mind doesn't add new commands — it reuses Forge and Council under the hood. The combined estimate is just both stages, summed:

# Estimate both stages up front (shown as one range), then accept once.
python3 scripts/codex_council.py estimate --topic "<idea>" --mode standard --type forge --token-budget compact
python3 scripts/codex_council.py estimate --topic "<idea>" --mode standard --token-budget compact
Cheap by design Two stages, one accepted estimate, compact by default. The handoff between them is just the proposal — never full transcripts — so a full run doesn't double your tokens.