The MCP ecosystem moves fast. New servers, new Claude Code skills, new agent frameworks every week. The distribution infrastructure for indie builders in that space is basically nonexistent — no curated channels, no automated submission pipelines, no recurring visibility mechanism. You ship, you post once, you move on.
marketing-pipeline fills that gap. One command onboards a project:
marketing onboard --name my-tool --repo owner/repo --kind mcp-server
It fetches the README, extracts the problem and key facts, and saves everything to projects.yml. After that, a GitHub Actions cron at 14:00 UTC on weekdays handles the rest: rotating posts across Bluesky, Mastodon, Dev.to, and Hashnode, and submitting to type-specific directories — MCP Registry, Smithery, Glama, and PulseMCP for mcp-server projects; awesome-claude-code for claude-skill projects.
The kind field is doing real work here. It routes each project to the directories that actually matter for that category rather than dumping everything into the same generic buckets.
One specific detail worth calling out: there's a hard antislop gate in pipeline/antislop.py that rejects posts before publication if they contain tokens like 'excited', 'game-changer', 'AI-powered', emoji, hashtags, or exclamation points. This matters because generated marketing copy defaults to that register and it reads as noise in developer communities. The gate isn't optional — it runs before every post.
The tradeoff I haven't automated away: awesome-claude-code requires human submission via their GitHub issue form. The pipeline generates the payload; you submit it once per project. That's the right constraint on their end.
Top comments (0)