Most MCP servers never get indexed anywhere useful. You push to GitHub, maybe post once on Bluesky, and six months later it has twelve stars from people who found it by accident.
The distribution problem isn't writing one good post — it's showing up repeatedly across the places developers actually look, and doing the directory work that puts your tool in front of someone running a search.
marketing-pipeline handles both. One onboard command fetches your README, extracts the problem and key facts via Claude, and saves everything to a config file:
marketing onboard --name my-tool --repo owner/repo --kind mcp-server
From that point, marketing cycle — which runs daily via GitHub Actions at 14:00 UTC on weekdays — rotates through your registered projects, picks the least-recently-used angle, drafts platform-appropriate posts (Bluesky at ≤300 chars, X at ≤280, Dev.to/Hashnode at 150–400 words), and publishes them.
For MCP servers specifically, the kind field triggers directory submissions to the official MCP Registry, Smithery, and GitHub Topics. The MCP Registry matters because Glama and PulseMCP pull from it automatically — one submission, three indexes.
The part I spent the most time on: a hard-reject gate in pipeline/antislop.py that kills posts containing specific tokens before anything gets published. The blocklist covers the obvious growth-marketing tokens (excited, game-changer, unlock, empower, AI-powered), plus emoji, hashtags, and exclamation points. Not because the platforms care — because posts that read like press releases don't get engagement from developers.
Requirements: an Anthropic API key, plus credentials for at least Bluesky, Dev.to, and Hashnode. Mastodon and Slack are optional.
Top comments (0)