Skip to content

New Skills — August 15, 2026 (Evening Sweep)

Fourth skills.sh sweep of August 15. 40 API queries collected 3,944 unique skills; cross-referenced against the hermes/ tree with the newest batch-page dirs excluded, then re-verified the entire carry-over queue on publisher pages. The dominant pattern again was queue underestimation: getsentry 20.7K → 85.3K, emblemcompany 8.8K → 78.6K, cloudai-x 7.9K → 80.0K, ljagiello → 71.6K (never counted before). Every queued candidate was re-verified per the stale-assessment rule before drafting or re-parking.

New Publisher Clusters — Guided This Sweep

Cluster Skills Installs Tier Guide
getsentry/skills 31 85.3K 🟢 Sentry Agent Skills Setup
cloudai-x/threejs-skills 10 80.0K 🟡 Three.js Agent Skills Setup
emblemcompany/agent-skills 9 78.6K 🟡 Emblem Company Agent Skills Setup
ljagiello/ctf-skills 12 71.6K 🟡 CTF Security Skills Setup
brightdata/skills 22 26.9K 🟢 Bright Data Agent Skills Setup
digitalsamba/claude-code-video-toolkit 13 14.5K 🟡 Claude Code Video Toolkit Setup
langfuse/skills 6 13.1K 🟢 Langfuse Agent Skills Setup
199-biotechnologies/claude-deep-research-skill 1 9.4K 🟡 Deep Research Skill Setup
mintlify/docs 5 9.2K 🟢 Mintlify Docs Skills Setup
lottiefiles/motion-design-skill 1 8.1K 🟢 Motion Design Skill Setup
ceorkm/mobile-app-ui-design 1 6.7K 🟢 Mobile App UI Design Skill Setup

Method Notes

  • 40 API queries → 3,944 unique skills → top-120 clusters all present in the tree → same-day triage recovery excluded new-aug15-2026* dirs → 5 not-guided candidates → carry-over queue re-verified on publisher pages (17 fetches) → 11 clusters drafted, 7 parked, 4 dropped.
  • Single-skill audit tiers: motion-design 🟢 (3/3 Pass), mobile-app-ui-design 🟢 (3/3 Pass), deep-research 🟡 (Snyk Warn). Multi-skill suites tiered by publisher: official vendor orgs (Sentry, Bright Data, Langfuse, Mintlify, LottieFiles) 🟢; community or crypto-adjacent publishers 🟡.
  • resend/react-email and resend/email-best-practices surfaced in the re-diff but were skipped — resend-skills-setup.md already covers the family.
  • Hot leaderboard (1H delta) added watch candidates: alibaba/open-code-review (4.3K), asksurf-ai/surf-skills (3.8K), streakyc/googleworkspacecli (1.7K), uni-stack/uniwind (5.3K).

Evaluated and Queued (next sweep)

Cluster Installs Reason parked
uni-stack/uniwind 5.3K RN tooling, at bar; watch next sweep
amazonappdev/devices-agent-skills 5.1K Fire TV migration niche
alibaba/open-code-review 4.3K Below bar; brand authority watch
asksurf-ai/surf-skills 3.8K Below bar
ningzimu/codex-ppt-skill 3.5K Below bar
dmmulroy/anti-slop 1.1K Below bar
zhaono1/agent-playbook 50.3K Top skill (self-improving-agent, 33.0K) already guided via charon-fan; remaining 23 skills ≤1.2K each

Dropped below minimum: sonilo-ai/skills (234), algolia/skills (235), tencentmusic/qqmusic-skills (714), jimliu/baocut (181).

Carry-over queue remains parked per playbook: herdr, last30days, agent-config, stop-slop, web-access.

Notable Signals for CorpusIQ

  • Sentry's suite is the strongest security + code-review skill cluster catalogued to date — a natural pairing for our GitHub work and the PR security gate.
  • Bright Data's competitive-intel, brand-listening, and seo-audit skills map one-to-one onto our research and SEO operations; scrape/search are a resilient fallback path for degraded search backends.
  • Langfuse tracing addresses the exact gap our multi-model operation has: attribute cost per model and debug agent loops.
  • The video toolkit (ffmpeg, remotion, elevenlabs) overlaps the same primitives our UGC pipeline runs on — a reference for pipeline hardening.
  • CTF skills give a structured offensive-security checklist for authorized testing, plus OSINT methodology we can reuse for research.

Index State After Sweep

  • catalog/index.md: +11 entries
  • marketplace/index.md: header 865 → 876, footer 916 → 927
  • last_updated on both indexes: 2026-08-15