If you have only heard the name “OpenClaw” in passing, here is what it is, why people care, and how Sired fits in—without assuming you already run a dev environment.
OpenClaw is an open-source way to run AI agents on your own machine: you install a command-line tool, give your agent a personality and tools (email, files, web search, and more, depending on setup), and chat with it locally or connect it to messaging apps. It exploded in popularity because it put a familiar pattern—“like ChatGPT, but yours, scriptable, and hackable”—into something developers could actually ship and iterate on, with a loud community sharing configs and skills.
Think of it as the “home lab” or “pro” lane for people who want full control: their disk, their keys, their workflows.
Sired is built for people who do not want to live in a terminal. You describe what you need; we host your Ally, connect channels (SMS, Slack, webhooks, and more), and handle billing and scaling. You get a working agent in minutes in the browser—not after reading a dozen READMEs.
That does not compete with OpenClaw; it complements it:
We offer an add-on that can suggest approaches before your agent acts and learn from outcomes after—so agents get smarter from shared experience, not just your single chat thread. You can use that from Sired-hosted allies or wire it into a custom or OpenClaw-based setup via our skill and APIs.
The rest of this page is for integration details. Jump to: CIP · Skill · Export · SDK · How to integrate
Enrich before each LLM call — suggested approaches, common mistakes, patterns. Ingest after each interaction — contribute outcomes so the collective learns. No Sired agent required: use any agentId as a label.
Drop-in skill for OpenClaw. Calls siredEnrich(task) before your LLM and siredIngest({ input, response, outcome }) after. One config block: API URL, key, agent ID.
sired cip setup provisions an org + API key and prints config for the skill. Auto-detects agent ID from package.json openclaw.agents or the current directory. Works with self-hosted or remote Sired.
Create agents in Sired, then export as an OpenClaw workspace (SOUL.md, openclaw.json, config). Run locally when you want full control.
Use enrich() and ingest() in custom agents. npm install sired-sdk or pip install sired-sdk. Works with LangChain, OpenClaw CLI, or any runtime.
Your agent learns from patterns across Sired users. Before each LLM call, enrich with suggested approaches and common mistakes. After each interaction, ingest so the collective improves.
Create agents in Sired (no terminal, no config). When you want full control, export to OpenClaw and run locally.
Sired allies don't just chat — they search the web, send email, create reports, run on schedules. Add more capabilities as you grow.
Add the sired-collective skill to your OpenClaw agent. Run sired cip setup to provision a key and get config, then paste into the skill.
Create an agent in Sired, then export it as an OpenClaw workspace. Download a zip with SOUL.md, openclaw.json, and all config. Run it on OpenClaw.
Export URL: GET /api/agents/:id/export (requires auth or setup token)
Use the Sired SDK in your custom agent. Two calls: enrich(task) before acting, ingest({ input, response, outcome }) after.
npm install sired-sdk · pip install sired-sdk