AIO Agent Company Kit — plain-text operating system for an AI agent workforce: charter, 9 role cards, QA gates, registers, receipts

YOUR AGENTS ARE TALENTED. THEY ARE ALSO UNMANAGED.

Every power user of OpenClaw-class CLIs, Claude Projects, custom GPTs, or

coding-agent tools hits the same wall: more agents, more chaos. Work vanishes

between sessions. Nothing is auditable. Two agents overwrite each other. Nobody

knows what "done" means. Human companies solved this centuries ago — with an org.

AIO Agent Company Kit is that org, generalized and ready to drop into any

instruction-friendly AI harness. It is not software: it is 26 plain-text files

that turn an agent pile into a company with law, lanes, records, and a quality

gate.

WHAT'S INSIDE (26 files, all readable text)

- org/CHARTER.md — the constitution: ownership, the session loop, records law

(local timestamps, append-only registers), honesty law, money protocol,

incident law

- org/ORG-CHART.md + ROUTING.md — nine lanes with reporting lines, escalation,

dispatch triggers, the handoff contract, and the anti-patterns that kill

multi-agent setups

- org/roles/ — NINE role cards (CEO, Research, Product Dev, Marketing, Support,

Finance, IT Ops, Quality Manager, Evolution), each paste-ready as a project's

instructions: mission, inputs/outputs, routing rules, boundaries, done

definition, QA checklist

- org/QA-GATES.md — four gates: evidence (validate before building), build

(three-way tie-outs, zero placeholders), publish (independent, veto-capable,

with scrub scan + honesty check), and the post-publish defect loop

- Five registers — pipeline BOARD, DECISION LOG, LEDGER, INCIDENTS (blameless,

5-Why/8D, closure only on verified fixes), and STATE working memory

- STANDUP + WEEKLY REVIEW cadences, PREREQUISITES with the harness

compatibility matrix, TOKENS reference (14 placeholder tokens, fill guidance)

- RECEIPTS.md — sanitized proof of what this system did for our studio in one

day: 4 products built, QA'd and published, with real tie-out numbers, its own

defects caught and fixed, and a clock bug the records standard caught in

itself. Our results, our models — no income claims, and day-one revenue was

$0, stated plainly.

WHY THIS ONE

- Registers as product: other kits ship roles; this ships the record-keeping

law (append-only, timestamped, one-writer) that makes an agent org auditable

at all.

- A quality system, not just prompts: an independent publish gate that can veto

the CEO, and a defect loop where every incident closes only on a new,

verified check.

- Receipts: evidence the system ships and self-corrects — the credibility

artifact no competitor includes.

- Harness-agnostic: works with OpenClaw-class agent CLIs, Claude Projects or

Claude Code, custom GPTs / ChatGPT Projects, ZCode-class coding CLIs, or any

tool that follows written instructions. Paste-and-point wiring, ~15 minutes.

HONEST NOTES (read before buying)

- You need an instruction-friendly AI harness you already run, your own API

access (your provider bills your usage), and a text editor. Nothing of that

is included — no code, no installer, no scripts, no keys, and nothing phones

home. Setup is 30–60 minutes (QUICKSTART walks every step).

- The kit organizes agent work; it promises no autonomous overnight output and

makes no income claims of any kind. The receipts are our studio's results,

not a forecast of yours.

- v1.0, first release, from a small agent studio that runs on this exact

system. Spotted a flaw? Message us and it gets fixed in the next free v1.x

update.

YOU GET

- AIO-Agent-Company-Kit-v1.zip — 26 plain-text files (charter, org chart,

routing, QA gates, 9 role cards, 5 register templates, 2 cadences,

prerequisites matrix, receipts, token reference, quickstart, license)

- Free updates to all v1.x versions