Sprout Generator
The generator IS the platform proof.
1Principles
ValuesBeliefsControls
Thesis dies without proof
- Platform claims 7 ventures, 0 run independently
- Manual wiring: 29 hours, 30% boot failure rate
- The generator exists. The test exists. Step 1.6 is the moment of truth.
| Problem | Question | Decision |
|---|---|---|
| Platform thesis is aspirational, not proven | Does one command produce a working venture? | Run the generator for berleytrails — binary pass/fail |
A mushroom pushing through cracked pavement — the mycelium underneath connects to six other sprouts waiting to emerge.
2Performance
ValuesBeliefsControls
Seven lines, no partial credit
| Measure | Target | Now |
|---|---|---|
| Generator runs | Exit 0 | Untested |
| Venture row exists | slug = berleytrails | Untested |
| Agent card registered | berleytrails namespace | Untested |
| WorkChart seeded | ventureId = berleytrails.id | Untested |
| A2A responds | HTTP 200 | Untested |
| Entities stubbed | 3 registered | Untested |
| drmg-sales unaffected | ventureId = null | Untested |
| Problem | Question | Decision |
|---|---|---|
| No measurement = no proof | Which of the 7 lines fails first? | Run acceptance test, fix in order |
An industrial control panel with 7 indicator lights — all currently dark, waiting for the first startup sequence.
3Platform
ValuesBeliefsControls
Eighteen tables, one command
| Layer | Tables | Status |
|---|---|---|
| Accounting | 5 | Scoped |
| Planning | 1 | Scoped |
| Value | 1 | Scoped |
| Venture | 11 | Scoped |
| WorkChart | 3 | Migration applied |
| Problem | Question | Decision |
|---|---|---|
| 18 tables scoped but never tested with a second app | Does the schema compose or couple? | Cold boot berleytrails against real DB |
An underground mycelium network — 18 glowing nodes across 5 distinct clusters, all connected to a central hub labeled 'venture'.
4Protocols
ValuesBeliefsControls
Step 1.6 unblocks everything
| Phase | What | Effort | Unblocks |
|---|---|---|---|
| Sprint 0 | Generator + test + runner | Done | First run |
| Sprint 1 | Run generator, boot, verify | 1 day | Domain entities |
| Sprint 2 | Domain tables + CRUD scaffold | 3 days | Real A2A |
| Sprint 3 | A2A execution + access control | 5 days | Second sprout |
| Sprint 4 | Second venture + cross-A2A | 2 days | Platform proof |
| Problem | Question | Decision |
|---|---|---|
| Single-thread critical path bottlenecked at Step 1.6 | What caused the codex crash? | Resume manually — run generator from terminal |
A domino chain — the first domino (labeled 1.6) is the only one still standing. Everything downstream is already set up, waiting.
5Players
ValuesBeliefsControls
One works. Two proves it.
| Job | Struggling Moment | Hidden Objection |
|---|---|---|
| Platform engineer creates venture | "I need to manually wire 18 tables, A2A, agents" | "What if the shared libs aren't actually shared?" |
| Venture founder validates idea | "I know my domain but can't build the infra" | "Will this platform lock me in?" |
| Problem | Question | Decision |
|---|---|---|
| One app might prove it works by accident. Need two. | Which venture runs second — prettymint or howzus? | Berleytrails first, decide second after Phase 1 acceptance |
Two mushrooms growing from the same underground network — berleytrails (with a fishing hook icon) and a second unmarked sprout emerging beside it.
1 / 5
Five Questions
- Can the mycelium grow a new venture in one command?
- Are the shared libs actually shared — or just well-organized monolith?
- What breaks when a second app composes the same infrastructure?
- Does additive beat subtractive when building from zero?
- When two ventures run, does the platform compound or fragment?