Brief-Me · IncredAgents Skills | One command. Six voices. You decide.

Stop reading walls of AI text. Get the brief.

Type /brief-me at the end of any working session. Six specialist voices read the entire conversation, argue it out until every one of them agrees, and hand you a single page of decisions — each with named options. The orchestrator assembles all of it. It gets no say in any of it.

One command. That's the whole interface.

No flags. No arguments. No settings file. The conversation you just had is the input — the skill reads all of it, so you never re-explain anything.

your agent — end of any session

# you just finished a long working session. type one thing:

> /brief-me

# six voices read the whole conversation, converge, then render

# your brief as its own clean page — cards, not a wall of chat text.

Six voices read your session. Then they have to agree.

Every brief is assembled from six required lenses. Each one looks at a different part of your work — and each verdict on the page names the voice that produced it, so nothing is anonymous.

The Researcher

Pattern recognition

Reads everything you found and names the connections across it — the pattern you were too close to see.

The Advisor

Synthesis

Thinks two weeks ahead. Second-order risks, what this decision sets in motion, what you'll wish you'd known.

Kaizen One

Consolidation

Looks for the waste — the friction this cuts, the steps it deletes, the drag it removes from every future session.

Kaizen Two

Risk

Asks the uncomfortable question: what breaks, and what does it cost to unbreak it?

The Linter

Correctness

Checks the work itself against the rules — violations, drift, anything that doesn't match the format you asked for.

The Builder

Implementation

Gets practical: which files, in what order, and what it costs in time to actually do the thing.

Split jobs. No off-the-record thinking.

Brief-Me splits the work in two: the orchestrator delegates, the voices do the thinking. Hard rules keep each side in its lane.

The Orchestrator Gets No Vote.

Its whole job is delegation — dispatch the six voices, collect their verdicts, assemble the page. It inserts no opinion, no ranking, no quiet pick of its own. If a voice didn't say it, the topic isn't on the page. Absent, never invented.

No Split Decisions Land On Your Desk.

If voices disagree, the dissenter gets re-dispatched with the others' arguments until every voice passes. A split tally is a status report dressed as a deliverable — you never see one. True deadlock? You get it as one question to break. Never a scoreboard.

Plain Speak. Real Numbers. Your Call.

Every card is one question in plain sentences — no jargon. Every option names the voice that proposed it, and "Other — I'll write my own" is always there. Names, numbers, and dollar amounts where they exist, because a fact you can check beats an adjective you can't.

Two minutes to install. One word to run.

Brief-Me ships inside the IncredAgents mod pack. One installer wires in the whole pack and asks only for the keys each skill needs.

your machine — three commands

$ git clone https://github.com/MyAiFreedomSystems/incredagents.git

$ cd incredagents

$ python3 onboard.py

# finish any working session, then tell your agent:

"/brief-me"

Just want this one skill? Grab the brief-me folder straight from the repo.