OpenClawCellCog

Your agents
just got a super-agent.

CellCog is the any-to-any execution layer for OpenClaw agents. Delegate research, documents, images, video, audio, and dashboards — no framework rewrites, no per-capability plumbing, just one API key.

Free API key · ClawHub skill or plugin install · Pay only for what you run · View pricing →

See it in action

One API key. Every modality. Zero framework lock-in.

Any OpenClaw agent can delegate to CellCog with three lines of Python. The SDK returns a unified response your agent narrates back — no streaming plumbing, no tool-specific wiring.

Python SDK
from cellcog import Client

client = Client(agent_provider="openclaw")
# reads CELLCOG_API_KEY from env

result = client.create_chat(
    prompt="Research quantum computing trends and generate a 2-page PDF",
    notify_session_key="agent:main:main",
    task_label="research",
)

print(result["message"])
# → "Research started. I'll notify you when the PDF is ready."
How it works

Three steps. Then your agents can do anything CellCog can.

No framework migration, no per-capability plumbing, no streaming handshake to implement. You add an API key. The rest happens.

01

Generate a CellCog API key

One key authenticates every OpenClaw agent that delegates to CellCog. Create it once, reuse across agents, environments, and deployments.

CellCog profile → API Keys
02

Ask your agent to install CellCog — two paths

ClawHub skill install is the lightweight option: the skill bundle teaches your agent every CellCog capability. OpenClaw plugin install gives you skills plus routing and setup bundled — richer integration when you want CellCog as a first-class part of your environment.

clawhub install cellcog · OR · openclaw plugins install clawhub:@cellcog/openclaw-plugin
03

Delegate from your agent

Your OpenClaw agent calls CellCog through the Python SDK. Two delivery modes — wait-for-completion for quick tasks, daemon-based notify-on-completion for long-running work. Automatic interim progress updates every 4 minutes so nothing feels stuck.

client.create_chat(prompt=…, notify_session_key=…)
Engineering

Built for agent developers, not platform tourists.

The hard parts — long-running task orchestration, streaming progress, capability discovery, SDK ergonomics — handled in code, not in documentation.

Any-to-any execution layer

One SDK call handles research, images, video, audio, PDFs, spreadsheets — no per-capability API design your agent has to learn. Every skill returns a unified response shape.

result = client.create_chat(
    prompt="…",
    notify_session_key="agent:main:main",
)
# Every skill: same unified response

Notify-on-completion, not poll-until-done

Long tasks don't block your agent. Start work with a notify_session_key; your daemon receives an event when CellCog finishes. No polling loops, no timeouts.

client.create_chat(
    prompt="Generate a 2-page PDF…",
    notify_session_key="agent:main:pdf_task",
)
# Agent continues. Daemon handles completion.

Progress updates every 4 minutes, automatically

Long-running work emits interim progress without you asking. Your agent can surface 'still working…' to the user without any CellCog-specific plumbing.

# CellCog emits AGENT_UPDATE events every 4 min.
# Your daemon handler receives them for free.

Auto-discovered skills catalog

Install once. CellCog grows new skills server-side. Your agent discovers them automatically — no breaking changes, no version bumps, no code edits.

# A year from now, CellCog adds a new skill.
# Your agent already knows how to use it.

Two install paths, same key

ClawHub skill install for lightweight capability extension. OpenClaw plugin install for full first-class integration. Both share the same API key and capability surface.

clawhub install cellcog
# OR
openclaw plugins install clawhub:@cellcog/openclaw-plugin
What OpenClaw agents delegate

Six real patterns. One execution layer.

Your agents stay focused on orchestration. CellCog handles the deliverables. Everything below is a working pattern on the OpenClaw network today.

Research pipeline → formatted PDF

Your research agent sources data. Delegate the write-up to CellCog. Get a 2-page PDF back with citations, charts, and formatting.

Content workflow → hero image

Your marketing agent drafts a blog post. Delegate the hero image with style + dimension specs. Get a ready-to-publish PNG or WebP.

Support ticket → explainer video

Your support agent handles a "how do I…" ticket. Delegate a 30-second explainer video to CellCog. Get a shareable MP4.

Financial analysis → Excel model

Your data agent pulls ticker data. Delegate the Excel model with formulas, charts, and formatted sheets. Get a .xlsx file.

Sales outreach → personalized deck

Your CRM agent identifies a prospect. Delegate a 5-slide personalized pitch deck. Get a .pptx file.

Ops agent → live dashboard

Your ops agent monitors metrics. Delegate an interactive HTML dashboard. Get a URL your team can share instantly.

Frequently asked

Questions we hear most.

One API key.
Every capability your agents need.

CellCog is the any-to-any execution layer already running inside the OpenClaw network. Add it to your agents in minutes.