Publish to social media from OpenClaw, step by step
Install the SocialCannon skill with one ClawHub command, add credentials privately, verify with a sandbox post, then let your OpenClaw agent draft and schedule with approval.
Updated · 6 min read
OpenClaw is good at turning a loose request into a finished task. With the SocialCannon skill installed, that task can be “turn these release notes into posts and schedule them for next week”. This guide installs the skill, connects it safely, and shows how to keep you in control of what gets published.
What the skill does
The skill gives your OpenClaw agent access to the SocialCannon API. It can draft and schedule posts for Facebook, Instagram, LinkedIn, TikTok, YouTube and X, and check what is queued. On Pro and above it can also read post analytics on Instagram, X and YouTube and reply to supported public comments. It works with the accounts you have connected to SocialCannon. It does not read or answer private direct messages.
Step 1: Install the skill
The skill is distributed through ClawHub as a community skill. From your OpenClaw workspace, with the ClawHub client installed, run:
clawhub install socialcannonRead the skill before you enable it, as you would any skill, and do not skip a security warning to get it working. Its listing is on ClawHub (opens in a new tab). Installing it does not connect anything yet; it only gives the agent the instructions and tools.
Step 2: Add your credentials privately
If you do not have a SocialCannon account yet, sign in with Google or GitHub; this creates a free account. In the dashboard, open Settings → API Keys and copy your client ID from the API Credentials card. If you already use your credentials elsewhere, reuse that secret: pressing Generate API Secret replaces it and breaks anything using the old one. Otherwise, generate one now (it is shown only once). Then put the two values into the skill’s environment or OpenClaw’s secret configuration:
SOCIALCANNON_CLIENT_IDSOCIALCANNON_CLIENT_SECRET
Step 3: Prove it works with the sandbox
Before anything touches a real account, ask the agent for a sandbox test. It should list your accounts first to confirm the credentials work; an empty list is fine, because no social account is connected yet. Then it should create exactly one post with accountId: "sandbox" and no scheduledAt, and report success only with the returned post ID and status. The sandbox simulates a publication; nothing is sent to a social network. If a request fails or its outcome is unclear, the agent should stop rather than try again. If you prefer to let the agent do the setup itself, paste this prompt:
Help me connect OpenClaw to SocialCannon so you can help with my social posts.
Read https://socialcannon.app/openclaw/setup.txt and guide me through setup one step at a time.
Keep my credentials out of this chat. Start with a sandbox test, and ask me before publishing or scheduling anything to real accounts.The prompt points the agent at the public plain-text setup guide, which tells it to keep credentials out of the chat, start with the sandbox, and ask before publishing.
Step 4: Connect your social accounts
Once the sandbox test passes, connect the real accounts. Social accounts are connected in a browser, because each network asks you to sign in and approve access yourself. Open the SocialCannon dashboard, go to Accounts, and connect the accounts you want the agent to use. The Free plan covers two accounts on Facebook, Instagram, LinkedIn, TikTok and YouTube; X needs a paid plan.
Step 5: Keep approval in the loop
The API does exactly what it is asked. A post without a scheduled time publishes immediately, and a scheduled post is published at or after its time; a busy queue or a network limit can delay it, and a post whose account has lost access is held until you reconnect and reschedule it. Approval therefore has to live in how you instruct the agent. A good routine:
- Ask for drafts first, for example “draft three posts from this blog article, one per network”.
- Edit or approve each draft in the conversation.
- Tell the agent which approved drafts to schedule, and when.
- Ask it to confirm what was scheduled, with the post IDs and times it received back.
Apply the same rule to replies, retries, deletions and disconnecting accounts. The setup guide already tells the agent to ask first; keep that instruction if you edit the skill.
Things to try
- “Turn this week’s changelog into a LinkedIn post and an X post, and show me both.”
- “What is scheduled for the next seven days, and which days are empty?”
- “Reschedule Thursday’s Facebook post to Friday at 10:00 UK time.”
- “Show me new comments on my last Instagram post.” (comment replies and analytics need Pro)
Next steps
The OpenClaw page has the install command and setup prompt ready to copy. If you use another assistant, the MCP server guide covers the same tools for any MCP client, and the API reference documents every endpoint the skill calls.
Frequently asked questions
- How do I install the SocialCannon skill?
- From your OpenClaw workspace, with the ClawHub client installed, run clawhub install socialcannon. Then add SOCIALCANNON_CLIENT_ID and SOCIALCANNON_CLIENT_SECRET to the skill's private environment.
- Does installing the skill connect my social accounts?
- No. Social accounts are connected in a browser through each network's own sign-in screen. The skill uses whatever accounts are connected to your SocialCannon account.
- Will OpenClaw post without asking?
- The setup guide tells your agent to ask before real publishing or scheduling. Keep that instruction in place. The API itself publishes immediately when no scheduledAt is given.
- Can the skill read or answer my direct messages?
- No. SocialCannon covers posts, scheduling, supported public comments and replies, analytics and content workflows. It does not read or answer private direct messages.
Start with a free account
Free includes two connected accounts, unlimited posts and scheduling, and a sandbox for testing. Some features in this guide need Pro; the pricing page lists what each plan includes.