2026-03-05 Session Log
indie-maker: added 4 missing agents, fixed I/O connections, built ideation flows, separated into standalone repo
indie-maker
Completed the full 29-day sprint agent system and separated indie-maker into its own repository.
Background
indie-maker is a sprint system that takes an idea from zero to a validated MVP in 29 days. It runs as a set of Claude Code skills — you type /indie-planner or /indie-analyst and a specialized agent walks you through that phase conversationally.
Before today, the system had 6 agents. A full process review revealed 4 gaps in the coverage.
Filling the Gaps: 4 New Agents
Phase 1.5 — indie-ux (Kai, UX Architect)
There was nothing between planning (“what to build”) and visual design (“colors and components”). The middle layer — “what screens exist and how they connect” — was missing.
Kai produces:
- Full user flow diagrams (every screen, every branch)
- Lo-fi text wireframes for 3–5 key screens
- Error / empty state / loading state specs
Reviews against Nielsen’s 10 usability heuristics and applies cognitive load reduction principles throughout.
Phase 5 — indie-launcher (Leo, Launch Strategist)
D7–D13 was a manual gap. The existing /launch-kit generates marketing copy, but copy and launch strategy are different things. Copy is what you say. Strategy is when, where, and how you go public.
Leo handles:
- Product Hunt submission strategy (timing, tagline, first comment)
- Pre-seeding — collecting upvote commitments before launch day
- Multi-channel stacking: BetaList, Show HN, AI directories, newsletters simultaneously
- 7-day Build-in-Public content calendar (Twitter/LinkedIn drafts ready to post)
- Launch Failure Plan B: if upvotes fall below 50, here are 4 concrete options
Phase 8+ — indie-growth (Gio, Growth Strategist)
After the D29 Kill/Go gate returns “Go,” there was no agent to handle what comes next. The previous setup had a single line pointing to /business-avengers, which breaks context continuity.
Gio’s core principle: retention before acquisition. If Day-7 return rate is below 25%, don’t spend energy bringing in new users — fix the product first. New users who leave immediately don’t compound.
After retention is healthy: test 3 channels from a pool of 19, then commit to 1.
Phase 9 — indie-retro (Sage, Retrospective Lead)
A Kill verdict previously ended with a 3-line checklist. Kill isn’t failure — it’s data. But data only compounds if it’s captured and carried into the next sprint.
Sage runs a 30-minute structured retrospective:
- Failure autopsy across 4 lenses: Product / Market / Execution / Timing
- Root cause vs. symptom: “not enough users” is a symptom, not a cause
- Full audit of every assumption made on D1 against what actually happened
- 3 next-sprint principles — specific behaviors, not abstract lessons
Fixing 3 Broken I/O Connections
Adding new agents exposed gaps in the existing document flow.
Gap 1: indie-analyst delivered Kill/Go analysis as conversation only — no file saved. indie-growth and indie-retro both need to read those metrics, but there was nothing to read.
→ Fixed: indie-analyst now saves kill-go-report.md immediately after the verdict (both Go and Kill paths).
Gap 2: indie-retro writes lessons.md, but indie-market-researcher never read it at the start of the next sprint. Previous failures could repeat unnoticed.
→ Fixed: indie-market-researcher now Globs lessons.md at startup and factors previous learnings into market analysis.
Gap 3: indie-backend didn’t read design-brief.md. DB table names and API endpoint names could be generated with no awareness of the product’s brand naming.
→ Fixed: design-brief.md added to indie-backend’s context load.
Adding Ideation Flows
The system previously assumed you already had an idea when you started. Two explicit paths now exist:
Path A (Research → Planning): No idea yet
- Run
/indie-market-researcher→ desire-based market analysis → 3 concrete product idea candidates (each with target user, differentiation angle, estimated ARPU, feasibility score) - Pick one → run
/indie-planner
Path B (Planning → Validate → Planning): Have an idea already
- Start
/indie-planner→ after Q2 (personal experience check), prompted: “Have you validated demand for this?” - If not: run
/indie-market-researcher --validate "your idea"(~5–8 min) - Returns: Google Trends trajectory + community discussion signals + willingness to pay + best acquisition channel → saves
research/demand-validation.md - indie-planner reads that file and continues with demand context loaded
Validation verdicts:
- 🟢 Validated — proceed
- 🟡 Weak Signal — run this specific test first, come back with results
- 🔴 No Signal — reconsider the angle or start over with
/indie-market-researcherdiscovery mode
Separated into Standalone Repo
With 11 agents, 13 knowledge documents, and a full sprint playbook, indie-maker had grown beyond a plugin. Extracted from the claude-ai-engineering monorepo into its own repository.
- New location:
~/Documents/Projects/indie-maker/ - GitHub: https://github.com/JayKim88/indie-maker
- All skill versions reset to v1.0.0
- CLAUDE.md rewritten: full agent table, both sprint paths, complete document flow diagram
- All 11 symlinks updated:
~/.claude/skills/[skill]→ new repo path - Removed
plugins/indie-maker/anddocs/indie-sprint-playbook.mdfrom the monorepo
Final Agent Map
| Agent | Phase | Role |
|---|---|---|
| indie-market-researcher (Max) | D-1 (optional) | Market research + idea candidates + demand validation |
| indie-planner (Reid) | D1 morning | Idea validation + PRD + Kill criteria |
| indie-ux (Kai) | D1 afternoon | User flows + wireframes |
| indie-designer (Vera) | D2 | Brand + landing design |
| indie-frontend (Rex) | D3–D6 | Next.js + UI implementation |
| indie-backend (Axel) | D3–D6 | Supabase + API + payments |
| indie-infra (Sam) | D3–D14 | Vercel deployment + QA |
| indie-launcher (Leo) | D7–D13 | Product Hunt + community launch strategy |
| indie-analyst (Nova) | D15–D29 | Kill/Go analysis |
| indie-growth (Gio) | D30+ (Go) | Growth experiments |
| indie-retro (Sage) | D29 (Kill) | Structured retrospective |
Next
- indie-retro: optional additions — competitive intelligence extraction, cross-sprint assumption pattern tracking
- indie-analyst: evaluate
knowledge/market-intelligence-guide.mdconnection to pre-launch stage analysis - indie-planner: evaluate
knowledge/founding-pm-guide.mddepth upgrade - GitHub: add README.md to indie-maker repo for public discoverability