Close the skills-distribution gap: list the catalog on skills.sh + HuggingFace

[jeremylongshore]

Why

Competitive analysis (2026-06-11) against Iris Wei's growth-playbook catalog (gingiris.tools/skills) surfaced a clear distribution asymmetry:

• We out-build her ~75× — 447 plugins / 3,023 skills / 425 npm packages vs. her ~38 skills.
• But we under-distribute. She lists every skill on HuggingFace (datasets) + skills.sh (npx skills add) — the two highest-velocity install surfaces, each with its own discovery/SEO/social graph — plus 4 languages, per-skill install-count social proof, and a Pro tier.
• We are confined to our own site + npm + /plugin marketplace add. We get zero benefit from the skills.sh / HuggingFace discovery graphs.

This is the highest-ROI, lowest-effort move from the analysis: meet installs where the velocity already is.

Scope (first child — claude-23jb)

List the marketplace's top skills on skills.sh and HuggingFace Skills.

• Pick the top 30–50 skills by npm download / quality grade
• Publish to skills.sh under our handle so npx skills add jeremylongshore/<slug> resolves
• Publish the same set as HuggingFace skill datasets
• Cross-link both surfaces back to tonsofskills.com
• Confirm per-skill install telemetry begins accruing

huggingface/skills already uses the .claude-plugin/marketplace.json pattern that's compatible with how we structure SKILL.md, so the porting cost is low.

Epic (claude-3f7u) — future children, not yet filed

Surface per-skill install counts as the card hook · ship a "start here" meta-router entry skill · bind operator credibility to flagship skills · stand up a Pro tier · localize the top 20 (CN/JA/KO).

Beads: claude-3f7u (epic), claude-23jb (this scope)

• Jeremy Longshore
  intentsolutions.io

[jeremylongshore]

Linked bead: claude-3f7u (source of truth). Use bd-sync to mirror updates.

[jeremylongshore]

Linked bead: claude-3f7u (source of truth). Use bd-sync to mirror updates.