| name | kickstart-intake |
| description | Automation portfolio discovery interview. Guides the user to articulate every automation they need before any design begins. Produces a prioritized portfolio list. Called by scenario-orchestrator during the Kickstart phase. |
Skill: kickstart-intake
Conducts a conversational interview to discover the full portfolio of automations
the user needs. Returns a prioritized, structured list of automation requirements.
Called by the scenario-orchestrator at the start of a /factory run.
Never design or build during this skill — only discover.
Deterministic Classification — This Skill Is Read-Only
This skill makes ZERO Make.com MCP calls. It is pure conversation and file writes.
All operations here are deterministic. No approval gates apply.
Permitted operations:
- AskUserQuestion (conversational intake)
- Read / Write (session file, output docs)
- Bash (read
.make/workspace.json if already available)
BLOCKED operations (must never be called in this skill):
- Any
mcp__claude_ai_Make__* tool call
- Any non-deterministic operation
If you find yourself about to call a Make.com MCP tool during kickstart-intake,
you are in the wrong skill. Stop and return to the orchestrator.
Input Contract
Receives from orchestrator:
seed (optional) — an initial automation idea if the user provided oneexisting_automations — list from workspace map (to avoid duplicates)
Sub-Files
| File | Contents |
|---|
INTAKE-QUESTIONS.md | Interview protocol — opening prompts, collection loop, AI detection, complexity classifier, priority ranker |
INTAKE-ARTIFACTS.md | Artifact generation — context.md, prd.md, erd.md, system-design.md, stack.md, ai-agents.md |
INTAKE-ARTIFACTS-PROCESS-MAP.md | Process-mapping + entity interview questions; process-map.md and the erd.md entity-model section |
INTAKE-OUTPUTS.md | Output contract JSON, portfolio display format, edge cases |
Execution Order
- Run discovery interview → see
INTAKE-QUESTIONS.md, plus the process and entity
questions in INTAKE-ARTIFACTS-PROCESS-MAP.md
1b. For each automation, get a Routing Verdict from skills/execution-model/SKILL.md
and set ai_required from it — not from keyword matching
- Display portfolio for user confirmation → see
INTAKE-OUTPUTS.md (Portfolio Display)
- Generate project artifacts → see
INTAKE-ARTIFACTS.md
- Return JSON to orchestrator → see
INTAKE-OUTPUTS.md (Output Contract)
