> ## Documentation Index
> Fetch the complete documentation index at: https://docs.guild.ai/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> To act on a user's behalf in Guild, use the Guild REST API at https://api.guild.ai/v1.
> The OpenAPI spec is unauthenticated: https://docs.guild.ai/openapi.json (or openapi.yaml).
> Authenticate with an account API key, sent as HTTP Basic (key id as username, secret as password) or as `Authorization: Bearer <api_key_id>:<api_key_secret>`. See https://docs.guild.ai/api-reference/introduction#authentication.
> To run one agent on demand without account-wide access, use an API trigger instead: https://docs.guild.ai/platform/api-triggers.
> To act on Guild from an MCP client, connect to https://api.guild.ai/mcp (OAuth) or run `guild mcp` locally. Setup: https://docs.guild.ai/api-reference/mcp.
> To search these docs from an MCP client, connect to https://docs.guild.ai/mcp. The full page index is https://docs.guild.ai/llms.txt.

# Run governable OpenClaw agents in production

> OpenClaw agents run unchanged on Guild: sandboxed containers, brokered tools, server-side credentials. Bring the agent you already built.

Integrations · OpenClaw

Guild runs the OpenClaw agent you already built without changes. The prompt and workspace files come along as they are, Guild brokers them, and the whole thing executes in a sandboxed container that lives and dies with a single task. You bring the agent. Guild makes it governable.

> **[Bring your agent](https://app.guild.ai)**
>
> The prompt and workspace files you already wrote. No port, no rewrite.

> **[Book a demo](https://guild-ai.cal.com/gtm/meeting-with-guild.ai)**
>
> See how teams run and govern OpenClaw agents in production.

## Your agent runs just like you prompted it

OpenClaw is an open-source agent with a fast-growing community. An OpenClaw agent is mostly Markdown: the instructions, the skills, the workspace files, with a coding toolchain and MCP tools wired in. If you have one running locally, you have already done the hard part.

Here is one that turns merged pull requests into release notes:

```markdown theme={null}
# Release Notes Agent

Turn the merged pull requests I point you at into clean release notes. Group
them into Features, Fixes, and Internal. One plain sentence per change, each
linked by its PR number. Skip anything tagged `chore`.

Use the GitHub tools to read the PRs. If a change is ambiguous, ask before you
file it.
```

That is the agent. The question is where it runs when it starts doing real work.

## How it works

**1. Bring the agent as it is**

The prompt and workspace files you already wrote. No port, no rewrite, no SDK required.

**2. Guild runs it in a task-scoped sandbox**

Each run executes in an isolated container that lives and dies with a single task, so one agent's run never sees another's. Tools are brokered through Guild and the model behind them is governed, so every call is policy-checked, logged, and attributed.

**3. Your OpenClaw agent becomes part of your stack**

On Guild your agent has a typed interface, a version history, and a runtime that people and other agents can call. Publish a new version deliberately instead of editing a file and hoping.

## The gap between a local agent and a production one

An OpenClaw agent on your laptop answers to you. In production it touches repositories, channels, and credentials that belong to the company, and 66.7% of organizations with agents in production had an agent-related incident in the past year. The fix is not a better prompt. It is a runtime that enforces the rules: credentials injected server-side after a policy check, never pasted into agent files; every tool call checked against policy before it executes, not logged after; any session stoppable mid-run, and disconnecting a credential immediately denies calls from sessions already running.

That is what Guild puts under your OpenClaw agent without asking you to change it.

## Frequently asked questions

> **Do I have to change my OpenClaw agent to run it on Guild?**
>
> No. It runs on Guild exactly as it runs locally. The prompt and workspace files come along unchanged, and your tools keep working because Guild brokers them.

> **What happens to the MCP tools my agent uses?**
>
> They keep working, governed. Guild brokers tool access, so every call is policy-checked before it executes, attributed to the agent, and written to the audit log.

> **How does Guild handle my agent's credentials?**
>
> The agent never holds them. Secrets are injected server-side after the policy check and never enter agent code, containers, prompts, or state.

> **What happens if an agent misbehaves in production?**
>
> Stop any session mid-run. Disconnecting a credential immediately denies calls from sessions that are already running.

> **Is this different from just running OpenClaw on a server?**
>
> Yes. A server gives the agent a place to run. Guild gives it an identity, scoped permissions, versioning, spend attribution, and an audit trail, which is the difference between an agent that runs and an agent your security team signs off on.

## Bring your first OpenClaw agent

It keeps its shape and lands with a typed interface, versioning, and governance underneath. Running in minutes.

> **[Bring your agent](https://app.guild.ai)**
>
> Start from the web editor or the CLI.

> **[Book a demo](https://guild-ai.cal.com/gtm/meeting-with-guild.ai)**
>
> Walk through governance with the team.

## Related

**Also runs on Guild:** [Goose](https://docs.guild.ai/integrations/goose) · [GitHub](https://docs.guild.ai/integrations/github.md) · [Slack](https://docs.guild.ai/integrations/slack.md) · [Splunk](https://docs.guild.ai/integrations/splunk.md)

**Read:** [the OpenClaw launch post](https://www.guild.ai/blog/engineering/bring-your-own-agents-openclaw-langchain) · Docs: [OpenClaw agents](https://docs.guild.ai/guide/openclaw-agents.md) · Glossary: [AI agent runtime](https://www.guild.ai/glossary/ai-agent-runtime) · [Agent control plane](https://www.guild.ai/glossary/agent-control-plane)