itropa

ITROPA

Name a human need. Go make coffee. Come back to a full research report with scored product ideas.

ITROPA is a Claude Code plugin that turns Claude into an autonomous innovation research engine. You say /itropa:go Belonging and Claude runs a 12-phase pipeline — web-grounded research, future industry discovery, mechanism extraction, business analysis, product concepts — all personalized to your builder profile. No configuration, no APIs, no prompting. Just results.

/itropa:go Status

 Starting full research pipeline on "Status"...

 Phase 0: Research Agents ✓ (3 agents dispatched in parallel)
 Phase 1: Intelligence ✓ (web + knowledge data consolidated)
 Phase 2: Industry Tree ✓ (5 future industries generated)
 Phase 3: Triage ✓ (Top 3 selected for your profile)
 Phase 4: Branch ✓ (#1 expanded into 3 sub-industries)
 Phase 5: Mechanisms ✓ (15 framework analyses)
 Phase 6: Deep Dive ✓ (3 industries scored)
 Phase 7: App Concepts ✓ (5 product ideas generated)
 Phase 8: Cross-Pollinate ✓ (top 2 combined, 5 hybrids)
 Phase 8.5: Digest ✓ (compact summary generated)
 Phase 8.7: Graph + Runs ✓ (connections updated)
 Phase 9: Report ✓

 Top Opportunities:
   1. Reputation Intelligence API (84/100) — BUILD
      Time to MVP: 2-4 weeks | $2-5k MRR
   2. Credibility Graph Networks (71/100) — EXPLORE
   3. Status-as-a-Service Platform (68/100) — EXPLORE

 Quick win: CredScore CLI (weekend, $500-1k MRR)
 Ambitious bet: ReputationOS (1-2 months, $5-10k MRR)

Install

claude plugin marketplace add Joncik91/itropa-plugin
claude plugin install itropa-plugin@itropa-marketplace

Use

mkdir my-research && cd my-research
claude

/itropa:setup              # One time — sets up workspace + builder profile
/itropa:constraints        # "I'm a senior dev, side project, want $5k MRR"
/itropa:go Belonging       # Full autonomous pipeline

Note: /itropa:setup creates the workspace structure and asks for your builder profile. The MCP search server installs its own dependencies automatically on first start. You only need to run setup once per workspace.

Then just talk:

"Go deeper on #2"
"Cross-pollinate those top two"
"What can I build this weekend?"
"Explore Safety next"
"Compare my Status runs"
"What connects Status and Belonging?"
/itropa:status

4 commands total (setup, go, status, constraints). Everything else is conversation.

What Makes This Different

Claude IS the engine. No external APIs, no wrapper scripts, no token costs beyond your Claude subscription. The plugin encodes ITROPA's 5-framework analysis methodology as skills — Claude does all the thinking.

Fully autonomous. You don't pick which industry to analyze or which framework to run. Claude triages, prioritizes, and chains 12 phases without asking. You review the output, not manage the process.

Personalized to you. Describe yourself once — tech stack, goals, risk tolerance, available time. Every score, every verdict, every product concept is filtered through your profile.

Persistent across sessions. Everything saves as JSON. Close Claude, come back next week, /itropa:status shows exactly where you left off. Digests (~1KB each) enable fast session resume.

v2.1.1 Features

• Web-grounded research — 3 parallel research agents combine Claude's knowledge with real web search. Market data, companies, funding, and pricing come from actual web sources, not training memory. Falls back gracefully to knowledge-only if web search is unavailable.
• Timestamped runs — Re-research a need without losing previous data. Each run gets its own dated directory.
• Research graph — Shared mechanisms and themes across needs are tracked automatically.
• MCP search — Full-text search across all research data using SQLite FTS5.
• Run comparison — Compare how your research evolved across multiple runs.
• Digest-first resume — New sessions load compact summaries (~1KB each) instead of reading all files.
• v1 migration — Existing v1 workspaces are automatically detected and can be migrated.

The Pipeline