cognis-operations

A real, opinionated blueprint for running a company where small human teams supervise factories of AI agents. This is how Cognis Digital operates — and a template you can fork.

• 🧠 OPERATING-MODEL.md — the 4-layer agentic operating model
• 🗺️ ORG.md — the new org chart (humans + agent factories + AI-native roles)
• 🤖 AGENTS.md — the agent registry (roles, scopes, owners)
• 🛡️ GOVERNANCE.md — human-in-the-loop, guardrails, audit & accountability
• 📚 SOURCES.md — McKinsey, Berkeley CMR, MIT Tech Review, Okta, NIST

Sample result format (illustrative values — run on your own data for real findings):

{
 "id": "123456",
 "name": "John Doe",
 "start_date": "2022-01-01T00:00:00Z",
 "end_date": "2022-12-31T23:59:59Z",
 "status": "IN_PROGRESS",
 "priority": "HIGH",
 "tags": ["tag1", "tag2"],
 "created_at": "2022-01-15T14:30:00Z",
 "updated_at": "2022-02-10T12:45:00Z"
}

This repo is a documentation blueprint (no CLI) — an operating model you fork and adapt.

1. Get the repo:

   git clone https://github.com/cognis-digital/cognis-operations && cd cognis-operations

2. Start with the operating model — read OPERATING-MODEL.md for the 4-layer agentic model (Governance -> Control -> Coordination -> Cognitive).
3. Map it to your org — adapt ORG.md (humans + agent factories + AI-native roles) and register your agents in AGENTS.md with roles, scopes, and owners.
4. Put guardrails in place — apply GOVERNANCE.md for human-in-the-loop approvals, audit, and accountability before letting agent factories run.
5. Keep it living in CI — commit your forked AGENTS.md / GOVERNANCE.md to your own repo and review them in PRs, so the agent registry and guardrails stay version-controlled as your fleet grows. Sources behind the model are in SOURCES.md.

flowchart TB
 G[Governance Layer<br/>policy · risk · accountability] --> C[Control Layer<br/>guardrails · approvals · audit]
 C --> O[Coordination Layer<br/>orchestration · routing · memory]
 O --> K[Cognitive Layer<br/>models · tools · skills]
 subgraph Humans
 CEO[Founder / Chief of Agents] --- POD[Outcome pods<br/>2-5 humans]
 end
 POD -->|supervise| O
 K --> FLEET[(uncensored-fleet + 195 tools)]

cognis-operations composes with the 300+ tool Cognis suite — JSON in/out and a shared OpenAI-compatible /v1 backbone. See INTEROP.md for the suite map, composition patterns, and reference stacks.

Forward cognis-operations's findings to STIX/MISP/Sigma/Splunk/Elastic/Slack/webhooks via cognis-connect. See INTEGRATIONS.md .

COCL v1.0 — see LICENSE.