tibursocampos/agent-dev-toolkit / core/skills/orchestrate-analyze
orchestrate-analyze
VerifiedOrchestrated Delivery O1: triage a feature, spawn conditional Agent specialists, write FEATURE.md + CONTINUITY + US/TS under features/NNN-slug/. No app code. Use when invoking /orchestrate-analyze.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 18.5 KB |
| reference.md | markdown | 1.8 KB |
| references/arch-confirm.md | markdown | 4.0 KB |
| references/boundaries-handoff.md | markdown | 3.5 KB |
| references/command.md | markdown | 3.5 KB |
| references/intent-classification.md | markdown | 5.2 KB |
| references/memory-bank-gate.md | markdown | 881 B |
| references/must-not.md | markdown | 4.4 KB |
| references/process-common.md | markdown | 3.8 KB |
| references/spawn-map.md | markdown | 3.7 KB |
| references/story-synthesis.md | markdown | 18.8 KB |
| references/triage.md | markdown | 5.0 KB |
references/spawn-map.md
View on GitHub## Flag -> specialist mapping
**Canonical source:** `skills/_shared/agents/ROSTER.md` (`needs_*` table). Keep this section as a short pointer - edit ROSTER when the map changes.
Spawn Agent **only** when `subagents=native` **and** ROSTER says so (`SPAWN.md`). If `subagents=none` or Agent unavailable → **fallback** **in-parent write** to `ANALYSIS/` / `ARCH/` / `SEC/` (never skip; never CONTINUITY substitute for those flags). Parallelize when multiple specialists apply; concurrent Agent cap **≤4** per `SPAWN.md`. Brownfield / impact-unclear -> `repo-analyst` **and** `architect` (mirror ARCH; skip style re-pick only) + `database` when persistence is in scope. Greenfield / no established style -> spawn `architect` and run **Architecture confirm gate**. ARCH is **not** optional when `needs_domain` or brownfield. Optional stage notes: `impact` / `risk` / `generate-story` prompts.
**Stacks are not roster roles** - do not spawn `react`/`dotnet` agents in O1. Route implementation later via `ROUTING.md`.
**Must not (specialists):** app code; org-only tooling unless the repo already uses it; invent APIs; expand to 40 agent files. `qa_checklist` = CONTINUITY/STORY only (no Agent).
---
## Process — Spawn Agent specialists
**SPAWN first:** load `SPAWN.md`; consult capability `subagents` (`native` \| `none`). Prefer Agent when `native`; if `subagents=none` or Agent unavailable → **fallback** **in-parent write** to `ANALYSIS/` / `ARCH/` / `SEC/` — **never skip** required folders; never substitute a CONTINUITY handoff note for `needs_api` / `needs_domain` / `needs_database` / `needs_security` / brownfield. Never hard-fail for missing Agent. Concurrent Agent cap **≤4** per `SPAWN.md` (wave if more; do not invent a new cap). `needs_frontend` / `needs_devops` stay CONTINUITY-only (no Agent).
Spawn an Agent subagent **only** when `subagents=native` **and** `ROSTER.md` canonical `needs_*` / brownfield rules say so. Load prompt from `skills/_shared/agents/prompts/`. The parent prompt requires the return shape in `refine-story/references/finding-format.md` (zero or more finding blocks; a summary is not the only product). Who is called does not change. When multiple specialists apply, spawn **in parallel** within the SPAWN ≤4 cap; if more flags apply, batch in waves of ≤4 or ask in the user chat language to run série.
**Model (`SUBAGENT-MODEL.md`):** omit Agent `model` by default — child uses the **same model as the parent session**. Do **not** conflate with a host auto model or Memory Bank policy `auto`. Ask about a premium slug **only** for very hard work per that contract; on **não** / silence, spawn without `model`. Never pick a costlier model alone.
| Signal (see ROSTER) | Specialist | Prompt |
|---------------------|------------|--------|
| `needs_api` or brownfield / impact unclear | `repo-analyst` → `ANALYSIS/` | `prompts/repo_analyst.md` |
| `needs_domain`, contract-heavy API, **greenfield**, or **brownfield** (mirror; skip style re-pick only) | `architect` → `ARCH/` | `prompts/architect.md` |
| `needs_database` or brownfield with persistence | `database` → `ARCH/` DB slice | `prompts/database.md` |
| `needs_security` | `security` → `SEC/` | `prompts/security.md` |
| `needs_frontend` | *(no O1 specialist)* - CONTINUITY note only; route at implement via `ROUTING.md` |
| `needs_devops` | short CONTINUITY note only (no Agent) | - |
| Story drafting aid | use `generate-story` patterns | `prompts/generate-story.md` |
| Optional stage notes | `impact` / `risk` | `prompts/impact.md`, `prompts/risk.md` |
Parent keeps lean context: synthesis + paths. Specialists must **not** write app code. Do **not** call `*-developer` to implement. `qa_checklist` is CONTINUITY/STORY only - never spawn an Agent for it.
See also § Flag -> specialist mapping.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 468k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |