model · serpro69 — Claude Code skill: domain glossary and a decision queue
model reads one bounded context and leaves a glossary, a traps page, and a queue of questions that need a human answer.
no license ★ 155 updated 2026-10-29
npx skills add serpro69/claude-toolbox --skill model -a claude-codeWhat model does
The skill maps a single bounded context: which concepts live in it, where they attach to code, and what about them still needs a person to decide. Output is two durable pages — a glossary and a list of divergences and traps — plus the result the author treats as primary, a queue of precise questions tagged by role.
Seven phases get you there. Scope intake fixes the context, the forcing question and whether a kit already exists. A gate then pulls the real asks out of linked tickets. Code reading follows state → time → invariants, keeping what the code proves apart from intent. After that come the draft, a verification pass that has to raise new questions instead of re-confirming, a self-check against the kit's own conventions, and a close that indexes the reasoning. It runs in the main session and writes files as it goes.
Instruction files load first, every time: model-process.md, archaeology.md, kit-contract.md, the four shared guards, the capy protocol. Skip that order and you get plausible vocabulary with none of the method behind it. It is not a behavioural code reviewer, and it will not write your feature design doc.
What it helps with
- Map a bounded context's vocabulary before design starts
- Record divergences and traps where domain terms clash
- Surface decidable questions for stakeholders, tagged by role
- Tie domain concepts to the code that implements them
- Refresh an existing kit with deltas instead of a rewrite
Example prompts for the agent
$kk:model: model the orders context and build the glossary- What are the concepts in this bounded context? I need a domain reference and the open questions
- Update the domain reference for billing — the linked tickets changed
How to install model
1Skills CLI (Vercel)
npx skills add serpro69/claude-toolbox --skill model -a claude-code2Manually: put the files in place
~/.claude/skills/model/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/model/
1Skills CLI (Vercel)
npx skills add serpro69/claude-toolbox --skill model -a codex2Manually: put the files in place
.agents/skills/model/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add serpro69/claude-toolbox --skill model -a cursor2Manually: put the files in place
.cursor/skills/model/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install serpro69/claude-toolbox kodex-plugin/skills/model2Manually: put the files in place
.github/skills/model/SKILL.md1Skills CLI (Vercel)
npx skills add serpro69/claude-toolbox --skill model -a gemini-cli2Platform CLI
gemini skills install https://github.com/serpro69/claude-toolbox --path kodex-plugin/skills/modelManually: put the files in place
.devin/skills/model/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about model
How do I install model in Claude Code?
Run npx skills add serpro69/claude-toolbox --skill model -a claude-code. The skill lands at ~/.claude/skills/model/SKILL.md.
What does model do?
It builds a domain-reference kit for a bounded context: a glossary and a page of divergences and traps. It also returns a role-tagged decision queue, and appends it to docs/feat/wip/<feature>/design.md when that folder exists.
How is model different from $kk:design?
model comes earlier and deals in concepts rather than a feature design. It is the first producer in the architecture flow, and $kk:design never calls it; $kk:review-architecture checks what it emits.
Similar skills
All skills for Agent setup: memory, context, orchestration →model by other authors comes first.