openclaw-prime-skill

Docs-first priming skill for OpenClaw operators.

Use it before admin work on an OpenClaw setup: gateway changes, channel debugging, remote access, security reviews, deployments, upgrades, model routing, or general operator triage.

What this repo contains

• SKILL.md — the skill definition and workflow
• references/admin-primer.md — condensed operator map
• agents/openai.yaml — optional prompt surface metadata

Install

Codex

Clone the repo:

git clone https://github.com/NicholasSpisak/openclaw-prime-skill.git ~/Projects/openclaw-prime-skill

Link it into your Codex skills directory:

mkdir -p ~/.codex/skills
ln -s ~/Projects/openclaw-prime-skill ~/.codex/skills/openclaw-prime

If a previous version exists:

trash ~/.codex/skills/openclaw-prime
ln -s ~/Projects/openclaw-prime-skill ~/.codex/skills/openclaw-prime

Restart the Codex session after linking so the skill list refreshes.

Other skill-based agent setups

Point your agent at this repo directory, or copy the contents into a skill folder where SKILL.md sits at the root.

Onboarding

1. Know the job

This skill is not a generic OpenClaw explainer. It is an operator primer.

Goal:

• refresh the live docs index
• reload the core gateway and architecture pages
• expand only into the task-relevant doc cluster
• separate docs truth from the actual environment
• produce a short operator brief before taking action

2. Know the doc flow

The skill starts from live docs, not memory.

Core sources:

• https://docs.openclaw.ai/llms.txt
• https://docs.openclaw.ai/sitemap.xml
• local primer: references/admin-primer.md

Core pages read every time:

• index.md
• start/hubs.md
• gateway/index.md
• gateway/configuration.md
• gateway/configuration-reference.md
• gateway/security/index.md
• gateway/remote.md
• gateway/troubleshooting.md
• cli/index.md
• channels/index.md
• concepts/architecture.md
• concepts/multi-agent.md

3. Invoke it

In a Codex session:

Use $openclaw-prime before we touch this gateway config.

Typical triggers:

• planning or executing gateway config changes
• debugging pairing, routing, or channel failures
• reviewing auth, proxy, tailscale, or secret handling
• validating deployments, upgrades, or migrations
• answering operator questions where stale docs would be risky

How to use it effectively

Best pattern

Ask the agent to use the skill first, then do the real task second.

Examples:

Use $openclaw-prime, then review this remote gateway setup and tell me the auth risks.

Use $openclaw-prime, then help me debug Slack channel routing.

Use $openclaw-prime, then plan the safest path to migrate this OpenClaw deployment.

What good output looks like

The skill should leave you with three things before deeper work starts:

1. concise OpenClaw architecture + ops summary
2. exact docs pages that govern the next action
3. local facts still needed before any recommendation or change

What it forces the agent to check

• deployment target
• config source
• auth mode
• exposed surfaces
• active channels and nodes
• secret boundaries
• sandbox and elevated-policy risk

This is the main value: it prevents docs-default assumptions from leaking into a live environment.

Repo workflow

Pull updates when the OpenClaw docs or the skill workflow changes:

cd ~/Projects/openclaw-prime-skill
git pull

If this repo is symlinked into ~/.codex/skills/openclaw-prime, no extra install step is needed after updates.

Source

This skill is built around the live OpenClaw docs at https://docs.openclaw.ai/.