Phase 1 — Interview → Plan — Productivity

Phase 1 of building a Claude Managed Agent: interviews the founder about the one job the agent should do, then produces a build sheet (CMA primitives table + v1/v2 deferrals + eval plan) without needing an API key yet. It walks six intake slots, maps answers to primitives, assembles a build-sheet.json, and validates limits.

agent-launcher claude-code

📂 View skill on GitHub ↗ 🚀 How to use

✨ What it does

Drives six intake slots (job, trigger, inputs, actions, definition-of-done, recurrence) via AskUserQuestion.
Maps answers to CMA primitives with interview_planner.py.
Assembles build-sheet.json with build_sheet_builder.py.
Validates limits with primitives_validator.py (PASS/WARN/FAIL).
Enforces hard rules: v0 core job only, deferrals for MCP and irreversible actions, no API key needed.
Provides forcing-question library with recommendations and citations.

🎯 When to use it

When the user says 'help me scope an agent' or 'I have an idea for an agent'.
When the user asks 'what should this agent be'.
When the orchestrator routes phase=interview.
When you need to plan an agent before any coding or API key setup.

🚀 How to use

Trigger by asking to scope an agent, e.g., 'help me scope an agent' or 'I have an idea for an agent'. The skill interviews you through six questions (job, trigger, inputs, actions, done, recurrence) and runs scripts to produce a build sheet. Example prompts:

help me scope an agent
I have an idea for an agent: triage overnight support email

📄 Output: A build-sheet.json file containing the primitives table, deferrals, and eval plan.

📦 Add this skill to Claude Code

# 1. Get the skills repo
git clone --depth 1 https://github.com/alirezarezvani/claude-skills /tmp/claude-skills

# 2. Copy this skill into your project (or ~/.claude/skills for all projects)
mkdir -p .claude/skills
cp -r /tmp/claude-skills/agent-launcher/skills/interview .claude/skills/interview

Skill source: agent-launcher/skills/interview/SKILL.md

⚠️ Good to know

Requires the user to provide clear answers to the six intake slots; the skill does not invent specifics and does not handle API keys (Phase 2).

❓ FAQ

Do I need an API key to use this skill?

No, the interview produces a plan; the key is a Phase-2 concern.

What happens if I have two jobs for the agent?

The skill recommends focusing on one job; two jobs should become two agents.

How are real integrations handled?

Real MCP servers are deferred to v1; v0 uses schema-true mock custom tools.

🤖 Overview, features, install steps and FAQ were generated from the project's SKILL.md on Sep 4, 2026. Always check the original source before running commands.