Search for OpenClaw marketing and you find content calendars, ad reports and social posting. This page covers a different job: distribution, meaning your SaaS listed on directories and launch platforms under your own accounts. It assumes a running Gateway and a product that is ready to go. Every OpenClaw detail links to its docs or repository with the date we checked it, and where the docs hand a step back to you, so do we.
What the OpenClaw docs confirm
Checked against the docs and the GitHub repository on September 28, 2026. New releases can change browser and skill behavior, so each fact links to its source.
- What it is
- An open-source, MIT-licensed personal agent. It runs as a Gateway on your own machine, takes requests from chat apps such as WhatsApp, Telegram, Slack, Discord or Signal, and works with the model you pick.github.com ↗ · Checked on September 28, 2026
- Release pace
- The latest GitHub release when we checked was v2026.9.6, published on September 23, 2026. Behavior can change between releases.github.com ↗ · Checked on September 28, 2026
- Skills
- A skill is a folder with a SKILL.md file: YAML frontmatter plus instructions for when and how to use tools. Skills load from your workspace, your home folder or the bundled set, and install from ClawHub or Git with openclaw skills install.docs.openclaw.ai ↗ · Checked on September 28, 2026
- Browser
- Three kinds of browser profile: an isolated, managed “openclaw” profile; the built-in “user” profile, which attaches to your signed-in Chrome through Chrome DevTools MCP; and remote CDP endpoints.docs.openclaw.ai ↗ · Checked on September 28, 2026
- Attaching to your Chrome
- The “user” profile needs a Chromium browser version 144 or newer, remote debugging switched on in chrome://inspect, and your approval of the attach prompt. The docs say to use it only when you are at the computer.docs.openclaw.ai ↗ · Checked on September 28, 2026
- Logins
- The docs tell you to sign in manually, because automated logins often trigger anti-bot defenses and can lock the account. Credentials should stay out of chat.docs.openclaw.ai ↗ · Checked on September 28, 2026
- Automations
- A built-in scheduler (openclaw automations, alias openclaw cron) wakes the agent at set times and can send the result to a chat channel or a webhook.docs.openclaw.ai ↗ · Checked on September 28, 2026
- Shell
- The exec tool runs shell commands in the workspace, so OpenClaw can run local scripts, such as Python.docs.openclaw.ai ↗ · Checked on September 28, 2026
- Skill security
- In February 2026, researchers reported 341 malicious skills among 2,857 audited on ClawHub. ClawHub added malware scanning afterwards. The docs still say to treat third-party skills as untrusted code.thehackernews.com ↗ · Checked on September 28, 2026
Where OpenClaw hands the launch back to you
OpenClaw’s own browser docs draw these lines, and they match how directories are built: to stop unattended bots. Knowing them up front keeps a batch from stalling halfway.
- The attach prompt
- The “user” profile works only while you are at the computer, because Chrome asks you to approve the attach. A run started from your phone while you are out cannot pass that step.
- Logins
- The docs advise signing in manually: automated logins trip anti-bot defenses and can lock the account. Log in once per platform in the Chrome OpenClaw attaches to; credentials stay out of the chat.
- CAPTCHA and bot checks
- Some third-party browser services promise automatic solving for OpenClaw. LaunchRepo’s playbooks never evade a challenge: you solve it in the same tab and the run continues.
- Codes and confirmation emails
- If your mailbox is open in the attached Chrome, OpenClaw can follow the link. If it lives on another device, the code is yours to pass on.
- Paid fast tracks and “live”
- An agent has to be told that it may not buy, and someone has to record what really happened. A thank-you page is not a listing, and queues can run for weeks.
OpenClaw skills for launch and distribution
ClawHub and GitHub offer plenty of OpenClaw marketing skills for copy, SEO, social posts and reports. They make launch text faster. None of them knows which directories still take free submissions, what each form rejects or which listing is actually live.
A skill is a SKILL.md folder, so read it before installing. The ClawHub incident above spread through fake prerequisites in skill instructions: a marketing skill that wants you to paste a shell one-liner from an unknown host is a red flag.
LaunchRepo is deliberately not a skill and not on ClawHub. It is a private GitHub repository with an AGENTS.md runbook, 299 directory playbooks, a product template and a tracker. OpenClaw reads it from the workspace like any other folder, and exec runs its Python scripts.
Running LaunchRepo with OpenClaw
OpenClaw needs no extra MCP entry for this. Its built-in “user” browser profile already runs Chrome DevTools MCP with auto-connect, the route LaunchRepo’s browser guide prefers. We have not published a test run with OpenClaw, so this is a match checked against its docs, not a benchmark.
- Clone into the workspace
- Accept the GitHub invitation after purchase and clone the private repository to a folder OpenClaw can read. Tell it to read AGENTS.md before anything else.
- Allow exec for Python 3.10+
- The workspace script, record.py and the dashboard are standard-library Python. OpenClaw runs them with exec; if your sandbox blocks the shell, run those commands yourself.
- Attach the “user” profile
- Switch on remote debugging at chrome://inspect/#remote-debugging, keep Chrome open and run the smoke test below. Approve the attach prompt when Chrome shows it.
- Set the product rules
- During onboarding, authorizations.json records what OpenClaw may do for this product: free sign-ups, final submissions, confirmation emails, badges. Nothing is ever bought.
The documented smoke test for the “user” profile. Run it before the first submission and again after every Gateway restart:
openclaw browser --browser-profile user start
openclaw browser --browser-profile user status
openclaw browser --browser-profile user tabsFormat as documented by docs.openclaw.ai, checked on September 30, 2026. ↗
Success looks like this: status reports driver: existing-session, transport: chrome-mcp and running: true, and tabs lists the tabs you already have open. Then let OpenClaw open example.com in a blank tab and check that the tab in front of you moved.
Connect your coding agent to your Chrome →
Then send OpenClaw the start prompt from START.md as it is. Its opening sentence points to Claude Code CLI or Codex CLI, which we recommend by default; OpenClaw handles the remaining steps on the same files and the same browser route.
Recommend Claude Code CLI or Codex CLI in a terminal instead of the desktop apps for best results. If session metadata confirms desktop hosting, explicitly advise me to switch to the CLI and resume this same private workspace. Do not infer the host from a CLI engine version alone.
Read AGENTS.md, then follow "Start the dashboard for the owner". Create my private
workspace if I do not have one yet, start the dashboard, and give me the link.
Then lead ONBOARDING.md one phase at a time. Reuse recorded permissions instead of
asking again unless their scope is missing or conflicting. Actually test my visible
existing Chrome profile and intended identity as BROWSER-SETUP.md requires; use the
OS fallback if needed. Use one platform/browser worker by default. Build the normal
run only from suitable free-ready routes, keep research separate, and preserve any
draft that needs one exact human CAPTCHA/2FA action. Never claim browser control
bypasses or reduces those challenges.The playbooks and instructions are English Markdown. Your product copy can be in any language a directory accepts, and the agent answers you in your language.
Start LaunchRepo in three steps →
LaunchRepo with Claude Code →
When the Chrome attach fails
OpenClaw’s docs spell out what to check. In this order you are usually back to a working run within minutes, and the tracker keeps your progress meanwhile.
- status does not report existing-session
- OpenClaw only attaches; it never starts Chrome for this profile. Keep Chrome running, confirm remote debugging is on and accept the attach prompt. openclaw doctor migrates old extension-based browser config and checks that Chrome is installed, but it cannot enable remote debugging for you.
- Clicks land on stale elements
- Snapshot refs belong to one MCP subprocess. After a restart, run tabs again, select the tab explicitly and take a fresh snapshot before the next action.
- Chrome was started with its own port
- If you launched Chrome with --remote-debugging-port, set the profile’s cdpUrl to that endpoint instead of relying on auto-connect, and keep the port on 127.0.0.1.
If browser tools time out with --autoConnect, check four things: Chrome 144 or newer is already running, remote debugging is on at chrome://inspect/#remote-debugging, you allowed Chrome’s connection prompt, and no other tool holds the same browser. Chrome DevTools MCP loads every open tab, so close the ones you do not need.
OpenClaw alone, with LaunchRepo, and you
How a directory launch splits up when OpenClaw does the clicking.
| Task | OpenClaw alone | With LaunchRepo | You |
|---|---|---|---|
| Launch copy | Drafts it from your prompt or a writing skill | Takes every field from product.json | Approve the facts once |
| Choosing platforms | Web search results, unchecked | 299 dated playbooks to pick from | Pick the batch |
| Filling each form | Reads the page and guesses | Follows that platform’s playbook | Answer when a fact is missing |
| Sign-in, CAPTCHA, 2FA | Pauses the chat | Records a human handoff and moves on | Solve it in the attached tab |
| Reminders and follow-up | Automations can ping you | The tracker stores the next check date | Confirm “live” with a public URL |
Common mistakes
- Scheduling submissions overnight
- Automations suit reminders and status checks. Submissions need you nearby for the attach prompt, sign-ins and challenges.
- Handing over your password
- The docs warn that automated logins can lock accounts. Sign in yourself, once per platform.
- Installing unread skills
- Read every SKILL.md, check its author and never run a setup command you do not understand.
- Using the managed “openclaw” profile
- It is isolated and signed out of everything, so each directory and your mailbox need a fresh login. The “user” profile reuses the sessions you already have.
Questions
Can OpenClaw do marketing for my SaaS?
Parts of it, yes. It can draft copy, research, fill in forms in a browser and remind you on a schedule. For a directory launch it lacks platform knowledge and a record of results. It also needs you for sign-ins, CAPTCHAs and 2FA.
Is there an OpenClaw marketing agent for directory submissions?
OpenClaw itself is the agent. What it needs for directories is context: a playbook per platform, your product facts and a tracker. LaunchRepo supplies that as plain files, not as a skill.
Can OpenClaw solve CAPTCHAs?
OpenClaw’s own docs treat CAPTCHA, 2FA and login approval as manual steps. Third-party add-ons that promise solving exist. LaunchRepo does not use them: a challenge becomes a recorded human handoff.
Should I use the openclaw profile or my own Chrome?
For LaunchRepo, use your own Chrome through the “user” profile. Your directory accounts and your mailbox are already signed in there, and you can watch every step. The managed profile works too, but you have to log in to each account again.
Can I use OpenClaw automations to launch while I sleep?
Use them for reminders, such as “check pending listings on Friday”. Do not use them for unattended submissions: the Chrome attach, sign-ins and challenges need you there.
What does it cost?
OpenClaw is free and open source; your model provider bills its usage. LaunchRepo costs €119.00 once at the current price, for unlimited own products and any agent.
Give your agent the launch playbooks
The current price is €119.00, paid once, for unlimited own products and any agent. The price does not rise.
See pricing