🆕 June 29, 2026 — 7 New Skills Discovered¶
Date: June 29, 2026 New Repos: 7 | New Skills: 7+ | Setup Guides: 7
Seven new community-contributed skills for Hermes Agent, discovered through GitHub search and skills.sh marketplace scanning. All include full setup guides in the catalog.
New Skills at a Glance¶
| # | Skill | Stars | Category | Setup Guide |
|---|---|---|---|---|
| 1 | coding-posture | 9★ | Agent Infrastructure | Setup → |
| 2 | ultimate-humanizer | 1★ | Content & Social | Setup → |
| 3 | clean-slate | — | Quality Assurance | Setup → |
| 4 | delegate-skills | — | Multi-Agent | Setup → |
| 5 | autolora | 1★ | Model Optimization | Setup → |
| 6 | hermes-whatsapp-secretary | — | Communication | Setup → |
| 7 | safari-web-agent | 1★ | Browser Automation | Setup → |
Skill Details¶
1. Coding Posture (9★)¶
Repo: alexei-led/coding-posture
A single SKILL.md that gives coding agents task-aware working modes — debug, fix, review, migrate, explain, and test. The agent adjusts its reasoning depth, tool selection, and output format based on the active posture. Works across Hermes, Claude Code, Codex, Cursor, and Pi.
Why it matters: Eliminates the need to write "be thorough" or "just give me a quick fix" in every prompt. The agent knows what mode it's in.
npx skills add alexei-led/coding-posture
2. Ultimate Humanizer (1★)¶
Repo: surdijon/ultimate-humanizer
Anti-AI-slop detection: 50 patterns, 2-pass self-audit, and 5-dimensional scoring (naturalness, variety, agency, specificity, rhythm). Optimized for French but works across languages. Includes Google spam-protection mode for SEO content.
Why it matters: Every AI-generated text carries patterns that readers — and search engines — increasingly detect. Ultimate Humanizer strips them systematically.
npx skills add surdijon/ultimate-humanizer
3. Clean Slate¶
Repo: sidhartha1s/clean-slate
Session-end verification that turns "looks done" into proven done. Audits every agent claim — merged PRs, passing tests, fixed bugs — against the real artifact. Produces a structured status report with ✅, ⚠️, and ❌ verdicts.
Why it matters: The #1 failure mode of coding agents is claiming completion without verification. Clean Slate makes verification automatic.
npx skills add sidhartha1s/clean-slate
4. Delegate Skills¶
Repo: bassemZohdy/delegate-skills
Spawn background agent workers in isolated git worktrees. The parent agent continues working while background workers handle long-running tasks — with self-healing monitoring that restarts failed workers.
Why it matters: True parallel agent work without context pollution. One agent handles the conversation while workers handle the heavy lifting.
npx skills add bassemZohdy/delegate-skills
5. AutoLoRA (1★)¶
Repo: DJLougen/autolora
Hermes fine-tunes itself. Curates successful agent traces, trains a LoRA adapter on RunPod via Unsloth, and hot-swaps the improved model — all while keeping costs capped via Stripe payment thresholds.
Why it matters: This is the self-improving agent loop made real. The agent gets better at your specific workflows over time.
npx skills add DJLougen/autolora
6. Hermes WhatsApp Secretary¶
Repo: ricktechmecha/hermes-whatsapp-secretary
Read, summarize, draft replies, and schedule WhatsApp messages — all through Hermes. Every reply requires your confirmation before sending. Priority filtering for important contacts.
Why it matters: WhatsApp is where business happens for many operators. This brings it into the agent workflow with appropriate safety gates.
npx skills add ricktechmecha/hermes-whatsapp-secretary
7. Safari Web Agent (1★)¶
Repo: treyxu23/safari-web-agent
Browser automation using your real Safari browser via native macOS CGEvent. Anti-bot bypass, login session persistence, and real browser fingerprint. Works on sites where Playwright gets detected.
Why it matters: Many business tools and dashboards block headless browsers. Safari Web Agent navigates them as a real user.
npx skills add treyxu23/safari-web-agent
Additional Repos Discovered (Curated for Future Review)¶
These repos were also found in the June 29 sweep. They're tracked for potential future setup guides:
| Repo | Description | Stars |
|---|---|---|
| Carojing/X-KOL-impressions | X/Twitter KOL outreach data extraction | 1★ |
| hanxing-AI/feynman-method-skill | Feynman Method interactive learning with hallucination defense | — |
| shagghiesuperstar/HERMES-Free-SOTA | FreeLLMAPI + Hermes MOA for SOTA at zero cost | — |
| julianargus01/agent-minds | Cognitive-function skills for agent thinking styles | 3★ |
| Jachinx-ai/deep-article-writing-skill | Chinese article writing: raw material to structured Markdown | 6★ |
| lunkerchen/skill-description-audit-skill | Audit and improve skill descriptions for trigger accuracy | — |
| jry21223/final-review-template-kit | LaTeX template + Hermes skill for exam review generation | 2★ |
| prasadmogulothu/agent-skills | Reusable agent skills with multi-model orchestration | — |
| hansai-art/hermes-agent-school | Failure patterns and diagnostic pack blueprint | — |
Stats¶
- Total marketplace skills: 464+ → 471+
- Categories covered this sweep: Agent Infrastructure, Content & Social, Quality Assurance, Multi-Agent, Model Optimization, Communication, Browser Automation
- Average stars: 1.7★ (early-stage community skills — high potential)
- Setup guides published: 7
← Skills Marketplace | Skills Catalog | Previous Update →
Powered by CorpusIQ