AgenticProductTeam: Instructions file for Claude Code

AgenticPod — Claude Code Instructions

AgenticPod is an agentic product lifecycle accelerator. You (Claude Code) are the UI. Users interact with you conversationally, and you drive the pipeline by calling library functions.

How It Works

AgenticPod runs a staged pipeline across four agent roles (PM → Research → Design → Engineering), producing structured artifacts (markdown + json) in ./artifacts/<project_id>/. Each stage calls an LLM to generate content based on prior artifacts.

Calling AgenticPod Functions

Because this is an ESM project with top-level await, use script files in ./scripts/ rather than npx tsx -e. Write a short .ts script, run it with npx tsx scripts/your-script.ts, then delete it after.

Example pattern:

// scripts/run-stage.ts
import { runStage } from '../src/index.js';
const r = await runStage('snackquest', 'pm');
console.log(JSON.stringify(r, null, 2));

npx tsx scripts/run-stage.ts

Commands (what users will ask you to do)

Start a new project

When the user describes a product idea, init a project:

import { initProject } from '../src/index.js';
const r = await initProject('THE IDEA TEXT', 'optional-project-id');
console.log(JSON.stringify(r, null, 2));

Then show the user the generated idea.json and ask if they want to refine it before proceeding.

Run the next stage

Run one stage at a time. After each stage, present:

1. What artifacts were created/updated
2. A brief summary of the content
3. Any quality issues
4. Ask: "Ready to proceed to [next stage]?"

import { runStage } from '../src/index.js';
const r = await runStage('PROJECT_ID', 'STAGE');
console.log(JSON.stringify(r, null, 2));

Stages in order: pm → research → design → engineering

Check project status

import { getProjectStatus } from '../src/index.js';
console.log(JSON.stringify(getProjectStatus('PROJECT_ID'), null, 2));

Run quality checks

import { runQualityChecks } from '../src/index.js';
console.log(JSON.stringify(runQualityChecks('PROJECT_ID'), null, 2));