skills-map
Skill233 starsChanged 7 months agoMITView on GitHub
Use at the START of any task, or whenever you are unsure which lich-skill applies. The router for this collection — it maps every lich-skill to the phase of work it belongs to (Decide → Spec → Plan → Build → Debug → Fan-out → Research → Media → Analyze) and walks a decision tree to pick the right one. Prevents the two failure modes of a skill collection: forgetting a skill exists, and reaching for the wrong skill at the wrong phase. Triggered by "/skills-map", "which skill should I use", "what skills do I have", or any moment of skill-selection doubt.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- AlpayC/claude-skill-settheir own version, first seen Sep 14, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- AlpayC/claude-skill-set★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository12 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- build-until-passSkill
- debug-hypothesisSkill
- frontend-designSkill
- go-no-goSkill
- google-analyticsSkill
- nano-bananaSkill
- spec-driven-devSkill
- spec-driven-dev-v2Skill
- subagent-briefSkill
- tavily-searchSkill
- wiki-aggregateSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 26
When a task spans multiple phases and you need the ordering.
- Distinctive: line 120
Route before you reach: When in doubt, run the decision tree before grabbing a tool — picking the wrong skill costs more than the 5 seconds …
- Very specific: line 121
go-no-gogates everything: If the task is "start X", that branch wins over every build/spec branch below it. - Distinctive: line 122
One phase, one skill: Don't stack skills for the same phase.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skills Map — the router for lich-skills
Most skills help you do the work. This one helps you pick which skill does the work.
A skill collection has a discovery problem: the skills only fire if the agent
remembers they exist and reaches for the right one at the right moment. Left to
chance, the agent forgets go-no-go and starts building, or hand-rolls a debug
crawl instead of running debug-hypothesis. skills-map is the index + decision
tree that closes that gap.
It does three things:
- Identify the phase of work the current task is in.
- Route to the matching skill via the decision tree.
- Sequence — most real work chains several skills; this gives the order.
When to Use
- At the start of any non-trivial task, before picking up tools.
- Whenever you catch yourself thinking "is there a skill for this?"
- When a task spans multiple phases and you need the ordering.
- On
/skills-map, "which skill", "what can you do", "list my skills".
When NOT to Use
- The task is a one-line answer or a trivial edit — just do it.
- You already know exactly which skill applies — invoke it directly.
- You are mid-skill — finish it; don't re-route every step.
The Map (by phase of work)
DECIDE SPEC / PLAN BUILD DEBUG FAN-OUT
┌────────┐ ┌──────────────┐ ┌────────────┐ ┌────────────┐ ┌──────────────┐
│go-no-go│──▶│spec-driven- │──▶│build-until-│─▶│debug- │ │subagent- │
│ Stage 0│ │dev / -v2 │ │pass │ │hypothesis │ │brief │
│ gate │ │ Spec→…→Ship │ │ red→green │ │ O→H→E→C │ │ ≤200w/agent │
└────────┘ └──────────────┘ └────────────┘ └────────────┘ └──────────────┘
RESEARCH MEDIA ANALYZE
┌──────────────┬──────────────┐ ┌────────────┐ ┌──────────────┐
│tavily-search │wiki-aggregate│ │nano-banana │ │google- │
│ web fetch │ N notes→pack │ │ text→image │ │analytics │
└──────────────┴──────────────┘ └────────────┘ └──────────────┘
DESIGN
┌──────────────┐
│frontend- │
│design │
└──────────────┘Decision Tree
Walk top to bottom; take the first branch that matches.
START
│
├─ About to start a NEW project / repo / contract? ───────────▶ go-no-go (Stage 0 — NO-GO is the default)
│
├─ Feature/refactor touching >1 file, needs a spec first?
│ ├─ hours–days, many files, multiple slices ─────────────▶ spec-driven-dev-v2
│ └─ single feature, normal scope ─────────────▶ spec-driven-dev
│
├─ A build / typecheck / lint / test is RED and you want green ─▶ build-until-pass
│
├─ Something is broken and you don't know WHY ────────────────▶ debug-hypothesis
│
├─ About to spawn subagents / Task / Agent fan-out ──────────▶ subagent-brief (compress FIRST)
│
├─ Building a UI / page / component ──────────────────────────▶ frontend-design
│
├─ Need facts / docs / citations from the web ───────────────▶ tavily-search
│
├─ Have N≥3 raw research notes to lift into one pack ─────────▶ wiki-aggregate
│
├─ Need an image generated or edited ─────────────────────────▶ nano-banana
│
├─ Question about site traffic / GA4 / SEO ──────────────────▶ google-analytics
│
└─ None match ───────────────────────────────────────────────▶ no skill; proceed manuallyQuick Reference
| Phase | Skill | Use it when… |
| Decide | go-no-go | Before any spec/plan/code — should this project start at all? NO-GO is the default. |
| Spec → Ship | spec-driven-dev | Any non-trivial feature/refactor touching >1 file. Gated Spec→Plan→Build→Test→Review→Ship. |
| Long-horizon | spec-driven-dev-v2 | Agent works hours–days across many files/slices. Project→Sprint→Task hierarchy + review loop. |
| Build | build-until-pass | A check is RED. Run → read error → smallest fix → re-run, until exit code 0 (cap 10). |
| Debug | debug-hypothesis | Bug with unknown cause. Scientific method: Observe → Hypothesize → Experiment → Conclude. |
| Fan-out | subagent-brief | Before spawning subagents. Compress each prompt to ≤200 words — no shared prefix cache. |
| Design | frontend-design | Building UI. Distinctive, production-grade interfaces that avoid generic AI-slop aesthetics. |
| Research | tavily-search | Real-time web search + content extraction via Tavily API. Facts, docs, citations. |
| Research | wiki-aggregate | N≥3 raw notes → one structured pack with cross-source claims + path:line provenance. |
| Media | nano-banana | Text-to-image or image edit via Nano Banana 2. 512 / 1K / 2K / 4K. |
| Analyze | google-analytics | GA4 traffic analysis + out-of-the-box SEO daily report for one property. |
Typical Sequences
Real work chains skills. The common pipelines:
New project go-no-go → spec-driven-dev → build-until-pass → (bug?) debug-hypothesis
Long migration go-no-go → spec-driven-dev-v2 → subagent-brief (per reviewer) → build-until-pass
Build a landing page spec-driven-dev → frontend-design → nano-banana (hero art) → build-until-pass
Research → report tavily-search → wiki-aggregate
Ship + measure …build-until-pass → google-analytics (post-launch SEO/traffic)Skill Rules
- Route before you reach. When in doubt, run the decision tree before grabbing a tool — picking the wrong skill costs more than the 5 seconds of routing.
go-no-gogates everything. If the task is "start X", that branch wins over every build/spec branch below it. Do not skip Stage 0 because you're eager to build.- One phase, one skill. Don't stack skills for the same phase. Finish the active skill before re-routing.
Anti-Rationalization
| Excuse | Reality |
| "I know the codebase, I'll just start coding." | That's the go-no-go / spec-driven-dev skip that produces the wrong thing fast. Route first. |
| "I'll debug it by reading around." | Unstructured crawling is the failure debug-hypothesis exists to prevent. Run the loop. |
| "I'll just spawn 10 subagents with full context." | One fan-out can burn a day of quota. subagent-brief first — always. |
| "No skill fits, I'll improvise." | Re-read the tree once. If it genuinely doesn't match, then proceed manually — and that's fine. |
Maintenance
This map is hand-curated. When a skill is added to or removed from this collection, update the phase map, decision tree, and quick-reference table here in the same change. A router that lists skills that don't exist — or omits ones that do — is worse than no router.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

