| name | orchestration/skill-navigator |
| trigger-words | ["orchestration/skill","navigator"] |
| description | Quick-reference navigator for all 169 active Hermes skills. Groups skills by domain-cluster (multi-agent, greyhack, mlops, creative, etc), tells you which skill to load for which task, flags clusters with overlap or gaps. Built from a real scan on 2026-07-02. Companion to the skill_view function โ load this FIRST to decide which skill to load next.
Trigger phrases: orchestration/skill, navigator.
Converted from Hermes-Skill (Yuno's mastery library).
|
๐ฅ Top-10 Most-Used Clusters (Decision Shortcuts)
โก NEW-USER INSTRUCTION: If you (Yuno or other agent) just loaded this skill, you probably need one of the clusters below. The cheat-sheet table maps user intents to primary + companion skills. If unsure between two skills, load the narrower one first (cheatsheet > domain > meta).
When a user request matches one of these patterns, load the bolded primary skill first:
| User Intent | Primary Skill | Companion Skills |
|---|
| Spawn parallel research/audit agents | multi-agent-pitfalls-cheatsheet | multi-agent-orchestration, multi-agent-work |
| Set up or debug Hermes itself | hermes-agent | hermes-maintenance, messaging-gateway-setup, hermes-admin |
| Configure Telegram/Discord gateway | messaging-gateway-setup | hermes-gateway, hermes-copilot-setup |
| Git workflow (commit, PR, branch, issues) | github-pr-workflow | github-issues, github-code-review, github-auth |
| Write/refactor/debug code | subagent-driven-development | plan, critic-gate, simplify-code, test-driven-development |
| GreyHack scripting | greyhack | greyhack-greyscript, greyhack-sandbox, greyhack-hermes-api |
| ML/Ollama/Local AI setup | ollama-local-hosting | mlops-suite, llama-cpp, local-ml-hosting |
| Build creative HTML/SVG/design | creative-suite | claude-design, popular-web-designs, html-artifact |
| System maintenance / cleanup | linux-system-maintenance | host-security-audit, local-ai-security-hygiene |
| Document writing / conversion | system-documentation | epub-export, pdf, obsidian, notion |
๐ Skill-Clusters (Domain-Grouped)
๐ Cluster 1: Multi-Agent & Orchestration (8 skills)
MULTI-AGENT CLUSTER โ when spawning parallel work
โโโ multi-agent-pitfalls-cheatsheet โ LOAD FIRST (trap-watchlist)
โโโ multi-agent-orchestration โ pattern + 5 phases
โโโ multi-agent-work โ end-to-end slash command
โโโ research-orchestration โ research-heavy plans
โโโ kanban-orchestrator / worker โ board-based long-running
โโโ hermes-agent โ general Hermes orchestration
โโโ the-dmz-transfer โ transfer from external source
โโโ depps-orchestration โ low-trust workers
When to load what:
- Just spawning subagents? โ
multi-agent-pitfalls-cheatsheet
- Full research project? โ
multi-agent-orchestration + cheatsheet
- One-shot research? โ
/multi-agent-work slash command
- Multi-day kanban board? โ
kanban-orchestrator
๐ ๏ธ Cluster 2: Software Development (26 skills โ LARGEST!)
CODE/DESIGN/TEST CLUSTER
โโโ Code-writing & patterns
โ โโโ subagent-driven-development โ plan โ delegate โ execute
โ โโโ plan โ write implementation plan
โ โโโ critic-gate โ output quality gate
โ โโโ simplify-code โ parallel cleanup agents
โ โโโ refactor / code-mod โ safe refactors
โ โโโ vanilla-js-tdz-helper-first โ JS helpers
โ
โโโ Code-review & QA
โ โโโ requesting-code-review โ pre-commit review
โ โโโ bash-script-audit โ bash safety scanner
โ โโโ security-code-checker โ LLM-generated-code risks
โ โโโ output-validator โ syntax pre-flight
โ โโโ systematic-debugging โ 4-phase root cause
โ
โโโ TDD & Testing
โ โโโ test-driven-development โ RED-GREEN-REFACTOR
โ โโโ python-debugpy โ remote debugger
โ โโโ node-inspect-debugger โ Node inspector
โ โโโ rag-pipeline-python โ RAG TDD example
โ
โโโ Language-specific
โ โโโ ki-murks-verhindern โ AI workflow gates
โ โโโ hermes-agent-skill-authoring โ writing Hermes SKILL.md
โ
โโโ Spike/Experiment
โ โโโ spike โ throwaway experiments
โ
โโโ Other
โโโ coding-skills-audit
โโโ multi-agent-research โ archived pattern source
๐จ Cluster 3: Creative & Design (23 skills)
CREATIVE CLUSTER
โโโ Suite / Orchestrator
โ โโโ creative-suite โ full UI-factory chain
โ
โโโ UI Components
โ โโโ ui-component-library โ buttons, inputs, cards
โ โโโ ui-color-system โ accessible palettes
โ โโโ ui-design-system โ full design tokens
โ โโโ ui-factory โ composite UI workflows
โ โโโ ui-dashboard โ data-schema โ KPI dashboard
โ
โโโ Web design & artifacts
โ โโโ claude-design โ one-off HTML artifacts
โ โโโ html-artifact โ self-contained HTML
โ โโโ popular-web-designs โ 54 real design systems
โ โโโ web-design-guidelines โ Vercel Web review
โ โโโ pretext โ @chenglou/p pretext browser
โ โโโ design-md โ Google DESIGN.md spec
โ โโโ sketch โ 2-3 variant mockups
โ
โโโ Diagrams & Visual
โ โโโ architecture-diagram โ dark-themed SVG
โ โโโ ascii-art โ pyfiglet, cowsay
โ โโโ ascii-video โ video โ ASCII
โ โโโ excalidraw โ hand-drawn diagrams
โ โโโ manim-video โ math animations
โ โโโ pixel-art โ NES/GB/PICO-8
โ
โโโ Other
โโโ humanizer โ strip AI-isms from text
โ๏ธ Cluster 4: DevOps & Maintenance (18 skills)
DEVOPS CLUSTER
โโโ Hermes maintenance
โ โโโ hermes-maintenance โ update side-effects
โ โโโ hermes-admin โ CLI/config troubleshooting
โ โโโ hermes-v7-sse โ V7 SSE package
โ โโโ hermes-v7-sse-server โ V7 SSE server
โ โโโ hermes-mcp-integration โ plugin-registry pattern
โ
โโโ Linux system
โ โโโ linux-system โ system troubleshooting
โ โโโ linux-system-maintenance โ safe cleanup CLI
โ โโโ linux-display-setup โ EDID / modelines
โ โโโ nvidia-laptop-gaming-tuning โ Optimus tuning
โ โโโ nvidia-suspend-resume โ GPU sleep/wake
โ โโโ docker-install-ubuntu โ Docker on Ubuntu
โ
โโโ Security
โ โโโ host-security-audit โ fwupd HSI / TPM / Secure
โ โโโ local-ai-security-hygiene โ secure local-AI setup
โ โโโ security-audit โ full security audit
โ โโโ system-security-audit โ ports/permissions audit
โ โโโ bash-script-audit โ script safety
โ
โโโ Messaging gateway
โโโ hermes-gateway โ gateway troubleshooting
โโโ messaging-gateway-setup โ platform setup
๐ฎ Cluster 5: Gaming (8 skills)
GAMING CLUSTER
โโโ GreyHack (the main cluster)
โ โโโ greyhack โ meta-skill + pipeline
โ โโโ greyhack-greyscript โ language reference
โ โโโ greyhack-sandbox โ local Python toolkit
โ โโโ greyhack-hermes-api โ in-game Hermes integration
โ โโโ greyhack-greyscript-deep-audit
โ โโโ greyscript-compiler-debugging
โ
โโโ Minecraft
โ โโโ minecraft-modpack-server
โ
โโโ Other
โ โโโ pokemon-player
โ โโโ teamspeak-server
โ โโโ game-library-management
๐ Cluster 6: GitHub Workflow (7 skills)
GITHUB CLUSTER
โโโ github-pr-workflow โ branch โ PR โ merge
โโโ github-issues โ CRUD via gh CLI
โโโ github-code-review โ PR review + comments
โโโ github-auth โ HTTPS/SSH setup
โโโ github-repo-management โ clone/fork/remote
โโโ codebase-inspection โ pygount stats
โโโ github-workflow โ full gh CLI workflow
๐ค Cluster 7: MLOps (11 skills)
MLOPS CLUSTER
โโโ Inference & Serving
โ โโโ serving-llms-vllm โ vLLM serving
โ โโโ llama-cpp โ GGUF local inference
โ โโโ ollama-local-hosting โ Ollama setup/manage
โ
โโโ Models
โ โโโ huggingface-hub โ HF model search/download
โ โโโ audiocraft-audio-generation โ MusicGen/AudioGen
โ โโโ segment-anything-model โ SAM zero-shot
โ
โโโ Research & Evaluation
โ โโโ dspy โ declarative LM programs
โ โโโ evaluating-llms-harness โ lm-eval-harness
โ โโโ llm-evaluation-troubleshooting
โ โโโ weights-and-biases โ experiment tracking
โ
โโโ Other
โโโ local-ml-hosting
โโโ obliteratus โ abliterate refusals
๐ฌ Cluster 8: Media (6 skills)
MEDIA CLUSTER
โโโ Audio
โ โโโ spotify โ Spotify CLI
โ โโโ heartmula โ song generation
โ โโโ songsee โ audio spectrograms
โ โโโ audiocraft-audio-generation โ also in mlops
โ
โโโ Video & Image
โ โโโ gif-search โ Tenor GIFs
โ โโโ media-tools โ YouTube transcripts etc.
โ
โโโ Other
โโโ youtube-content โ YT transcript โ blog
๐ค Cluster 9: Autonomous AI Agents (6 skills)
AUTONOMOUS-AGENTS CLUSTER
โโโ claude-code โ delegate to Claude Code CLI
โโโ codex โ delegate to OpenAI Codex
โโโ opencode โ delegate to OpenCode
โโโ coding-agents โ general coding agent orchestrator
โโโ hermes-agent โ (also in orchestration)
โโโ the-dmz-transfer
๐ Cluster 10: Productivity (16 skills)
PRODUCTIVITY CLUSTER
โโโ Daily Operations
โ โโโ daily-briefing โ system status briefing
โ โโโ petdex โ animated mascots
โ โโโ teams-meeting-pipeline โ Teams summary
โ
โโโ Documents
โ โโโ epub-export โ MD/PDF โ EPUB
โ โโโ pdf โ PDF reading
โ โโโ nano-pdf โ PDF editing
โ โโโ ocr-and-documents โ extract text from scans
โ โโโ powerpoint โ .pptx manipulation
โ
โโโ External Services
โ โโโ airtable โ REST API via curl
โ โโโ google-workspace โ Gmail/Drive/Docs/Sheets
โ โโโ linear โ issues/projects
โ โโโ maps โ OSM geocoding
โ โโโ notion โ Notion API
โ
โโโ Note-Taking
โ โโโ obsidian โ Obsidian vault
โ โโโ system-documentation โ MD doc tree
โ
โโโ Email
โโโ himalaya โ IMAP/SMTP CLI
๐จ Edge-Case Singletons (1-skill categories)
These are alone in their category โ load directly when relevant:
| Skill | Use When |
|---|
apple-notes / apple-reminders | macOS-only Apple ecosystem access |
arxiv | arXiv paper search |
architecture-diagram (root) | SVG cloud/infra diagrams |
ascii-video | Convert video/audio to ASCII |
blogwatcher | RSS/feed monitoring |
comfyui | ComfyUI local AI workflows |
computer-use | Desktop automation (background) |
context-mode | Context window optimization |
copilot-cli | GitHub Copilot CLI delegation |
data-science | Jupyter kernels |
debugging-hermes-tui-commands | Hermes TUI slash-command debugging |
dogfood | Exploratory QA of web apps |
excalidraw | Hand-drawn Excalidraw JSON |
firecrawl-web | Web scraping + screenshots |
find-skills | Discover/install new skills from skills.sh |
gif-search | Tenor GIF search |
google-oauth-setup | Google Workspace API OAuth setup |
ideation | Generate project ideas |
image-generate (root) | AI image generation |
kanban-codex-lane | Kanban + Codex CLI integration |
linux-display-setup | Custom EDID/modelines |
mcp-server-authoring | Build custom MCP servers |
mnemosyne-memory-provider | Mnemosyne native memory |
model-selector | Compare LLM models on Nous Portal |
multi-agent-orchestration | (also in orchestration cluster) |
native-mcp | Built-in MCP client |
notion | Notion API |
ollama-local-hosting | (also in mlops cluster) |
pdf-anthropic | PDF reading via Anthropic |
pokemon-player | Play Pokemon via headless emulator |
python-debugpy | Python remote debugging |
rag-pipeline-python | RAG pipeline (Tutor-style) |
red-teaming-godmode | LLM jailbreak testing |
research-orchestration | (in orchestration cluster) |
session-handoff | Model-switch handoff doc |
setup-guides-browser | Browser setup walkthroughs |
skill-install-workflow | skills.sh skill installer workflow |
songwriting-and-ai-music | Songwriting + Suno AI |
subagent-driven-development | (in software-development cluster) |
system-documentation | (in productivity cluster) |
telegram-clarification-prompt | Yuno's Telegram decision routing |
touchdesigner-mcp | TouchDesigner control |
voice-assistant-bots | Voice bot architecture |
web-archive-research | Common Crawl + WARC query |
web-design-guidelines | (in creative cluster) |
yuanbao | Yuanbao groups |
yuno-cleaner | Yuno Cleaner safe cleanup |
๐ Compliance Summary
Scan results (2026-07-02):
| Issue | Count | Severity | Action |
|---|
/home/bratan/ hardcoded paths | 21 files | LOW (cosmetic) | Optional: replace with ~/ |
| CLI drift (hermes skill/tools singular) | 0 | โ | Clean |
| Malformed frontmatter | 0 | โ | Clean |
| Read errors | 0 | โ | Clean |
The 21 hardcoded-path files (all in skill examples, not in code paths):
- gaming/greyhack, gaming/greyhack-greyscript, gaming/greyhack-hermes-api
- software-development/hermes-agent-skill-authoring, software-development/bash-script-audit, software-development/subagent-driven-development
- devops/hermes-maintenance, devops/local-ai-security-hygiene
- research/web-archive-research, research/research-tools
- autonomous-ai-agents/claude-code, autonomous-ai-agents/codex
- productivity/daily-briefing, productivity/google-workspace
- mlops/huggingface-hub
- orchestration/multi-agent-orchestration
- And a few others
Recommendation: Not worth fixing โ these are intentional examples showing actual user paths for pedagogical clarity. Skill consumers should mentally substitute ~.
๐ฏ Load-Order Heuristics (when in doubt)
- Trap-detector first โ
multi-agent-pitfalls-cheatsheet for any subagent work
- Pattern-source second โ
multi-agent-orchestration for the full pattern
- Domain-specific third โ e.g.
greyhack for GreyHack tasks
- Specialty-fourth โ e.g.
greyhack-greyscript for deep language reference
- Tool/script-specific fifth โ e.g.
ci-build.sh for build pipeline questions
If you're not sure between two skills: load the one with the narrower scope first (cheatsheet > domain > meta).
๐
When to re-scan this Navigator
Re-run the compliance scan + re-build this cheatsheet when:
- New Hermes version adds/removes skills (check
hermes skills list)
- User adds a custom skill
- Major skill consolidation (like 2026-07-02 multi-agent-research โ cheatsheet)
Re-scan from the reusable script
python3 ~/.hermes/skills/orchestration/skill-navigator/scripts/skill-scan.py
Quick-check (just counts)
python3 -c "import pathlib; print(sum(1 for f in pathlib.Path('/home/bratan/.hermes/skills').rglob('SKILL.md') if '.archive' not in f.parts and 'node_modules' not in str(f)), 'active skills')"
๐ Cross-Domain Navigation (Domain โ Method Linking)
Pattern discovered 2026-07-02: Skills in domain clusters (gaming, devops, mlops) should cross-reference skills in method clusters (orchestration, software-development) to enable tool-discovery chains.
Why it matters
A user working on GreyHack (gaming/greyhack) who needs subagent support should not have to know that multi-agent-pitfalls-cheatsheet lives in orchestration/. The domain skill should surface the method skill via a "Related Skills" or "Cross-Cluster Navigation" section.
When to add cross-references
| Condition | Action |
|---|
| Domain skill describes a workflow that could use subagents | Add ref to multi-agent-pitfalls-cheatsheet + multi-agent-orchestration |
| Domain skill mentions parallel research/audit | Add ref to research-orchestration |
| Domain skill involves code review/QA | Add ref to requesting-code-review or systematic-debugging |
| Domain skill references a meta-navigator | Add ref to skill-navigator (this skill) |
| Method skill lists use-cases for multiple domains | Add refs to relevant domain skills in its "See Also" section |
Format template (add to target SKILL.md)
## ๐งญ Related Skills (Cross-Cluster Navigation)
Skills that support this [domain] cluster but live elsewhere:
- **`skill-navigator`** (orchestration/) โ Meta-Navigator. Load FIRST when deciding which skill applies.
- **`multi-agent-pitfalls-cheatsheet`** (orchestration/) โ TRIGGER-WATCHLIST for `delegate_task` calls. Load BEFORE spawning subagents.
- **`multi-agent-orchestration`** (orchestration/) โ The 3-Expert Research PATTERN.
**[Domain] โ Orchestration Workflow:** For any [task type], load cheatsheet + orchestration together. They prevent the 4 HIGH-FREQUENCY pitfalls.
Workflow for adding cross-refs to a domain skill (proven 2026-07-02)
skill_view(name='<domain-skill>') โ read the SKILL.md to find the right insertion point- If skill has a "References", "Support Files", or "Related" section at the end โ append the
## ๐งญ Related Skills block after it
- If skill has no "References"/"Related" section โ append at end-of-file
- Verify: re-read SKILL.md to confirm the new section is well-positioned
- No frontmatter changes โ cross-refs are SKILL.md-body content only
Applied to (2026-07-02): all 4 GreyHack skills (greyhack, greyhack-greyscript, greyhack-sandbox, greyhack-hermes-api) plus bidirectional linking from multi-agent-pitfalls-cheatsheet โ skill-navigator.
Pitfalls
- Don't cross-ref every skill to every other skill โ that creates noise. Only link when a concrete workflow benefit exists (e.g. "GreyHack audit needs subagent traps").
- Keep the section at the END of the SKILL.md, not near the top โ cross-refs are secondary navigation, not primary purpose.
- One format per skill cluster โ all domain skills in the same cluster should use the same cross-ref header and format for consistency.
- Bidirectional when both loaded โ if skill A references skill B but B doesn't reference A back, that's OK for domainโmethod links (method skills serve multiple domains). But methodโmethod links (cheatsheetโnavigator) should be bidirectional.
๐ Final Note from Yuno
This navigator is a map, not the territory. Always verify a skill exists with skills_list or skill_view before recommending it. If a cluster has overlap (e.g. multi-agent-orchestration AND multi-agent-pitfalls-cheatsheet both about subagents), that's by design โ different abstraction levels serve different needs.
Lerneffekt: 1 Cheatsheet spart 30 Skill-Scan-API-Calls. ๐โ
Converted from Hermes-Skill by Yuno. Original SKILL.md preserved in references/original-hermes.md.
