OpenClaw users can now connect directly to Cencori using the normal OpenAI-compatible flow:

OPENAI_BASE_URL
OPENAI_API_KEY
optional CENCORI_AGENT_ID

No extra protocol, no custom gateway glue, no required X-Agent-ID header.

If your tool can target an OpenAI-compatible endpoint, it can use Cencori.

Why This Matters

OpenClaw is a strong agent runtime for desktop automation. Cencori adds the production layer around it:

centralized model routing
configurable system prompts
live monitoring
safety approvals with Shadow Mode
usage and cost tracking

So you keep OpenClaw’s autonomy and add control from one dashboard.

Quick Start

1. Deploy OpenClaw from Agent Marketplace

In Cencori:

go to Agents -> Agent Marketplace
deploy the OpenClaw blueprint
open the agent’s Configuration tab

2. Generate your Cencori API key

Create or roll a key in the same Configuration page.

3. Connect OpenClaw using environment variables

Codetext

export OPENAI_BASE_URL=https://cencori.com/api/v1
export OPENAI_API_KEY=cake_your_key_here
# Optional: force a specific agent (multi-agent setups)
export CENCORI_AGENT_ID=your_agent_id

Then run OpenClaw:

Codetext

openclaw onboard

4. Optional: use OpenClaw provider config file

If you prefer ~/.openclaw/openclaw.json:

Codetext

{
  "models": {
    "providers": {
      "cencori": {
        "baseUrl": "https://cencori.com/api/v1",
        "apiKey": "cake_your_key_here",
        "api": "openai-completions"
      }
    }
  }
}

What Happens in Cencori

Once connected:

all requests route through your Cencori gateway
selected model in Agent Configuration is applied
system prompt is injected from dashboard config
actions appear in Live Feed
Shadow Mode can require approval before risky tool execution

Important Behavior

CENCORI_AGENT_ID is optional.

By default, Cencori resolves the agent from the API key.
Set CENCORI_AGENT_ID only when you want explicit routing in multi-agent workflows.

Recommended First Run

For safety, start with:

Shadow Mode on
a test API key
a constrained prompt (limited directories/domains)

This gives you fast validation with guardrails before broader rollout.

Start Here

OpenClaw docs: /docs/agents/openclaw
Agents overview: /docs/agents/overview

If you're already using OpenClaw, point it to Cencori and you’re live.