Skills.sh Sweep - August 12, 2026 (Cron Update)¶
Automated discovery sweep across the skills.sh marketplace. Cross-referenced the full 215-skill nousresearch/hermes-agent catalog plus external marketplace listings against the existing 380+ entry catalog at corpusiq-docs/hermes/skills/catalog/.
Result: 7 skills not yet cataloged. All 7 setup guides drafted.
🆕 New Skills Discovered (7)¶
| Skill | Installs | Source | Setup Guide |
|---|---|---|---|
heartmula |
224 | nousresearch/hermes-agent | heartmula-setup.md ✍️ |
simplify-code |
187 | nousresearch/hermes-agent | simplify-code-setup.md ✍️ |
ideation |
111 | nousresearch/hermes-agent | ideation-setup.md ✍️ |
subagent-driven-development |
88 | nousresearch/hermes-agent | subagent-driven-development-setup.md ✍️ |
linear |
80 | nousresearch/hermes-agent | linear-setup.md ✍️ |
webhook-subscriptions |
80 | nousresearch/hermes-agent | webhook-subscriptions-setup.md ✍️ |
grounded-citations |
22 | nousresearch/hermes-agent | grounded-citations-setup.md ✍️ |
Skill Details¶
1. heartmula (224 installs)¶
Open-source AI music generation with HeartMuLa models (Apache 2.0). Four-model pipeline: HeartMuLa (3B/7B music LM), HeartCodec (12.5Hz audio codec), HeartTranscriptor (Whisper-based lyrics), HeartCLAP (audio-text alignment). Suno/Udio alternative - local, offline, no API keys.
2. simplify-code (187 installs)¶
Parallel code review with four focused reviewers running concurrently: Reuse (DRY), Quality (maintainability), Efficiency (performance), Altitude (architecture). Cleanup pass, not bug hunt - removes duplication, flattens complexity, cuts waste. Each reviewer runs in parallel.
3. ideation (111 installs)¶
Constraint-driven creative ideation. Generates 3 concrete project ideas from a creative constraint library, then builds the chosen one. Philosophy: constraint + direction = creativity. Works for code, art, hardware, tools.
4. subagent-driven-development (88 installs)¶
Multi-agent development workflow. Dispatches fresh subagents per task with systematic two-stage review (spec compliance then quality). Clean context per task, automated gating between steps. Consumes plan skill output.
5. linear (80 installs)¶
Linear.app integration via GraphQL API using curl. No MCP server, no OAuth - just an API key. Manage issues, projects, sprints, and team workloads from Hermes. Both UUID and short IDs (ENG-123) work.
6. webhook-subscriptions (80 installs)¶
External service webhook triggers for Hermes. GitHub pushes, Stripe events, CI/CD completions become agent triggers. HMAC signature verification, per-subscription routing, dynamic subscription management.
7. grounded-citations (22 installs)¶
Perplexity-style inline citations with verifiable source chains. Ledger-owned url → [n] mapping, verbatim quote verification, [unverified] flagging for model knowledge. verify --evidence command for fact-checking. Integrates with research-paper-writing.
📊 Existing Skills Verified (200+)¶
Full cross-reference against the 215-skill nousresearch/hermes-agent catalog confirmed the remaining 200+ skills are already cataloged, including all top-20 by installs: dogfood (5.4K), humanizer (740), hermes-tweet (695), popular-web-designs (563), yuanbao (561), hermes-agent (533), powerpoint (475), google-workspace (428), arxiv (424), research-paper-writing (397), claude-design (392), llm-wiki (371), jupyter-live-kernel (356), excalidraw (345), ascii-art (343), youtube-content (337), imessage (332), design-md (332), songwriting-and-ai-music (325), architecture-diagram (322).
🔍 Methodology¶
- 4 search queries across skills.sh CLI:
hermes(20 results)hermes agent(20 results)hermes skill(20 results)- Full
nousresearch/hermes-agentpage scrape (215 skills) - Cross-referenced against 380+ existing catalog entries
- 7 setup guides drafted for all uncataloged skills
- Sweep report published to
marketplace/new-aug12-2026/
📈 Ecosystem Growth¶
| Metric | Aug 10 (Pass 2) | Aug 11 | Aug 12 | Delta |
|---|---|---|---|---|
| Catalog entries | 358+ | 363+ | 370+ | +7 |
| Official skills tracked | 34 | 36 | 43 | +7 |
| Guides drafted this sweep | - | 5 | 7 | +7 |
| nousresearch/hermes-agent coverage | ~170/215 | ~175/215 | ~182/215 | +7 |
Next sweep: Automated cron - next run.
Powered by CorpusIQ
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.