Skip to content

106 social media skills for AI agents - strategy, writing, video, design, platform growth, publishing, and analytics. Works with Claude, Cursor, OpenClaw, Hermes & 40+ agents.

2.8 KB
Raw Blame

Contributing

Contributions are welcome — the two highest-value PRs are a new tool integration and a new skill. Both have a house pattern; matching it is most of the review.

The fastest useful PR: add a tool integration

  1. Write tools/integrations/<tool-name>.md — the connection layer: how to connect/authenticate, what the API/tool actually offers, pricing/plan/licensing gotchas. Study tools/integrations/kling.md for the shape.
  2. Add a row to tools/REGISTRY.md in the right cluster.
  3. Volatile facts (pricing, tiers, legal terms) must be attributed and carry a verify-quarterly note — never state them as timeless.

Adding a skill

Every skill is a directory under skills/ with this exact structure:

skills/<skill-name>/
  SKILL.md              # frontmatter (name matches the directory) + the craft
  references/           # 3–5 deep-dive files the agent loads on demand
  evals/evals.json      # test cases

House rules the review will hold you to:

  • One skill = one job-to-be-done, with explicit boundaries and routing to sibling skills. The frontmatter description is the trigger surface: write the phrases real users say, and route near-miss requests to the right sibling by name.
  • Every backticked skill-name mention must resolve to a real directory under skills/.
  • Opinionated, current, honest. Teach how top practitioners actually do the job now — not generic best-practice lists. No fabricated stats; attribute volatile figures and tag them verify-quarterly. No engagement-bait tactics, ever.
  • Publishing ground truth: skills hand finished content to a scheduling bridge whose capability surface is documented in tools/integrations/woopsocial.md — don't contradict that guide. Notably: it publishes and schedules, stores lifetime post analytics for TikTok, LinkedIn, Facebook, Instagram, Pinterest, YouTube, and Bluesky, and does not generate media. Measurement language: read WoopSocial post analytics for those networks; native dashboards for everything else. Never fabricate a number.
  • Nothing posts without explicit user confirmation. Side-effect skills keep the safety contract intact.

Before you open a PR

./scripts/validate-skills.sh

CI runs the same script; a red validator is an automatic revision request. If you add a skill that belongs in a topic pack, add it to scripts/packs.json too (the validator checks pack integrity).

Scope

This repo covers organic social media for AI agents. Out of scope: paid-ads campaign management, platform-API client libraries, and anything requiring credentials to be committed. When in doubt, open an issue first — happy to scope it with you.