The idea
These aren't tutorials you follow step by step in a terminal. They're instructions your OpenClaw agent can follow with you.
You copy a post, paste it into your chat with your agent, and say something like "let's set this up." Your agent reads the post, understands the goal, and walks through it with you — asking for credentials when it needs them, creating files, running commands, and explaining what it's doing along the way.
You're collaborating, not copy-pasting commands blind.
How it works
Every post has a "Copy article" button at the top. Click it and the full post is copied to your clipboard as clean text — no HTML, no formatting noise. Just the content your agent needs.
Then go to wherever you talk to your agent — Discord, Telegram, Signal, whatever channel you use — paste the article, and tell your agent to set it up.
That's it. Your agent handles the rest, stopping to ask when it needs input from you (like an API key or a preference).
Why this way
Most technical guides assume you're the one running commands. But if you're using OpenClaw, your agent is the one doing the work. It just needs to know what to do.
These posts are written with that in mind. They explain the "what" and "why" clearly enough for an agent to understand the goal, and include the specific commands and config so it can execute without guessing.
Your job is to approve what it does, provide secrets it can't access on its own, and make decisions when there's a choice to make.
A typical flow
- Find a post that solves something you want
- Click "Copy article"
- Paste it into your agent chat
- Say "set this up" (or whatever feels natural)
- Your agent reads it, starts working, and asks you questions as needed
- Done — the setup is live and you understood every step because you were part of it
If your chat has a character limit
Some channels — Discord without Nitro, for example — cap messages at 2,000 characters. Most articles are longer than that. Here are two ways around it.
Option 1: Use the OpenClaw CLI
Run a single command from your terminal. The article goes straight to your agent with no character limit.
openclaw agent --message "$(pbpaste)"
That sends whatever's on your clipboard. On Linux, swap pbpaste for xclip -selection clipboard -o. Or just pass the text directly:
openclaw agent --message "Paste the full article here"
Add --agent <name> if you have more than one agent configured.
Option 2: Use the Gateway HTTP endpoint
If your Gateway has the OpenAI-compatible endpoint enabled (it's off by default — run openclaw config set gateway.openai.enabled true to turn it on), you can send the article as an API call:
curl http://localhost:18789/v1/chat/completions \
-H "Authorization: Bearer YOUR_GATEWAY_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"model": "openclaw/default",
"messages": [{"role": "user", "content": "PASTE_ARTICLE_HERE"}]
}'
Replace localhost:18789 with your Gateway address, and YOUR_GATEWAY_TOKEN with your auth token. The agent processes the full article just like a normal chat message.
Both methods skip any channel character limits entirely — the article goes directly to your agent's context window.