hyperbuildlisted

Turns ONE app idea into a fully researched, designed, planned, and implemented application via the hyperbuild 19-step, two-stage pipeline: Stage A — PLAN (steps 1–12, autonomous: market recon, social mining, adversarial research audit — every load-bearing claim fact-checked by its own agent into a four-phase research archive — PRD, feature specs, stack research, 3 complete design systems with HTML mockups of every screen, adversarial VISUAL QA of every rendered screen, project-specific generated skills, epics/tasks) → ONE human checkpoint (`/hyperbuild-choose <a|b|c>`) → Stage B — BUILD (steps 13–16, autonomous: scaffold, implement, adversarial review, ship gate). This entry skill is a ROUTER. It contains no step procedures — it tells you which Skill to invoke for each step, in order, and how to recover a crashed run. Each step's instructions live in its own skill file (hyperbuild-1-intake through hyperbuild-16-ship-gate) and are loaded fresh into context when invoked.
zakariaf/hyperbuild · ★ 0 · Web & Frontend · score 72

Install: claude install-skill zakariaf/hyperbuild

# hyperbuild — multi-skill chain orchestrator You are the orchestrator. Your entire job in this conversation is: 1. Read this file once at the start. 2. Bootstrap (below) — or recover an existing run (see Recovery). 3. Invoke each step skill in sequence via the `Skill` tool. 4. Between steps, do nothing except mark todos and (optionally) think to `runs/<run_tag>/temp/orchestrator-notes.md`. You do NOT do the work of any step yourself. The step skills do. You just sequence them. --- ## How the chain works (READ THIS CAREFULLY) Each pipeline step is its own skill file. To run a step: ``` Skill(skill: "hyperbuild-N-stepname") ``` When you invoke a Skill, that skill's full procedure is loaded into your context **fresh**. You execute that step's procedure, hit its exit criteria, and return to this entry skill to invoke the next step. **Why this design?** Context rot. hyperresearch's V8 lesson: one 1200-line skill loaded once gets compacted away mid-run — by the time a late step needed its procedure, the orchestrator had forgotten it and silently improvised. Per-step skills loaded fresh at the moment they're needed survive compaction, because each step is self-contained and re-derives its inputs from disk. **The step skills** (all prefixed `hyperbuild-`): ### STAGE A — PLAN (autonomous, starts at `/hyperbuild <idea>`) | # | Skill name | What it does | Spawns (parallel) | |---|---|---|---| | 1 | `hyperbuild-1-intake` | Verbatim idea → `idea.md`; mint run_tag; resolve