LaunchRepo

You buy access to the repository and clone it. A guided onboarding step fills in a product.json context file with your name, description, categories, pricing and links, which becomes the single source the agent draws on for every form.

You then point your coding agent at a platform playbook. The agent reads the catalog entry, works out what that directory requires, and completes the submission using your product context. When it hits something it cannot do, a CAPTCHA or a human verification step, the stop rules tell it to hand control back rather than guess.

Outcomes are tracked in the repository as it goes, so you can see which of the 340 directories accepted the submission, which are pending and which need a manual pass.

• •Per-platform playbooks - individual instructions for each directory rather than one generic prompt, which is what makes the submissions actually complete.
• •Platform catalog - the list of 340+ launch directories with what each one requires.
• •product.json context template - your product described once, reused for every submission.
• •Stop rules - explicit instructions for handing CAPTCHAs and human checks back to you instead of failing silently.
• •Submission tracker - a record of what was submitted where and what happened.
• •Agent-agnostic - written for Claude Code, Codex and Cursor rather than tied to one vendor.

Solo founders and small teams shipping a SaaS who know directory submissions are worth doing and cannot face doing them by hand. It suits people who already pay for a coding agent and would rather point it at a defined task than write the prompts themselves.

It is a poor fit if you do not use an AI coding agent, because there is nothing here that runs on its own. It is also not a managed service: nobody submits anything for you, and the quality of the result depends on the agent you bring. If you want someone else to do the launching, this is the wrong shape of product entirely.