Agent runtime

AgentSEO + OpenClaw

Need a controlled SEO tool layer inside OpenClaw? Install the AgentSEO plugin, allowlist only the tools you want, and keep the runtime scoped instead of handing an agent a grab bag of external calls.

From a social thread? Copy the setup first, validate one live workflow, then decide whether it belongs in docs, automation, or your agent runtime.

First-party pluginTool allowlistsACP-compatible
Copy OpenClaw setupGet free API key

Why this page converts

Best for teams running real agents in production and caring about narrow permissions, attribution, and repeatable workflows.

Core workflows

What this page should convert

  • Attach only the SEO tools an OpenClaw agent needs for a specific workflow.
  • Run local audit, content gap, AI overview, and rank checks inside an agent runtime.
  • Group usage by projectId and workflowId in the plugin configuration.

Setup snapshot

OpenClaw plugin install

openclaw plugins install @agentseo/openclaw-plugin --pin

Call this a first-party AgentSEO plugin path. That is materially stronger than the n8n and Make positioning without implying OpenClaw-maintainer endorsement.

Implementation plan

How to set it up

  1. 1

    Install the published @agentseo/openclaw-plugin package.

  2. 2

    Add apiKey, apiUrl, and optional attribution defaults in the OpenClaw config.

  3. 3

    Allowlist only the AgentSEO tools the target agent should use.

  4. 4

    Restart the OpenClaw gateway and run a small smoke test before broader rollout.

Output shape

What users get back

  • Tool registration that matches OpenClaw's optional-tool pattern.
  • Direct access to core AgentSEO workflows from inside the runtime.
  • Cleaner governance because each agent can receive a scoped allowlist.

Limits

What to say plainly

  • Best for teams already committed to OpenClaw.
  • Requires runtime-level configuration, not just an API call.
  • Page copy should avoid implying that every endpoint is pre-allowlisted by default.

Visual flow

OpenClaw runtime shape

The operational win is scoped access: one agent, a narrow allowlist, and explicit polling for long-running checks.

1

Agent

OpenClaw agent receives a narrow SEO brief.

2

Plugin

First-party AgentSEO plugin exposes only allowlisted tools.

3

Job Status

Agent polls queued work through job status.

4

Controlled Output

Result returns without exposing broader runtime access.

Concrete example

Scoped SEO agent in OpenClaw

Create an OpenClaw agent with a narrow allowlist so it can run content-gap and rank checks without getting broad tool access.

Agent allowlist config

{
  "agents": {
    "list": [
      {
        "id": "seo",
        "tools": {
          "allow": [
            "agentseo_content_gap",
            "agentseo_rank_track",
            "agentseo_job_status"
          ]
        }
      }
    ]
  }
}

Runtime behavior

Agent can:
- enqueue rank checks
- inspect content gaps
- poll job status

Agent cannot:
- call broader tools outside the allowlist
  • OpenClaw is strongest when each agent gets a narrow operational role.
  • Use the plugin defaults for apiKey, projectId, and workflowId so every call stays attributable.
  • Run a smoke test after restart before widening the tool allowlist.

FAQ

Questions this page should answer fast

Package-level docs

Why feature OpenClaw early?

Because AgentSEO already ships a published first-party plugin, which gives the page both SEO depth and product credibility.

When is the ACP adapter better than the plugin?

Use the ACP adapter when the runtime is ACP-compatible but not OpenClaw itself.