$npx -y skills add aj-geddes/claude-code-bmad-skills --skill bmad-helpOrchestration spine and "what do I do next?" router for the BMAD Planning & Orchestrator plugin. Detects current planning state by scanning the output folder for which artifacts exist (project-context.md, decision-log.md, product-brief, prd/tech-spec, architecture, ux, epics, sto
| 1 | # BMAD Help — Orchestration Spine |
| 2 | |
| 3 | A lean router. It answers one question: **what do I run next?** It detects planning |
| 4 | state, infers the phase and track, and points to the next skill. It is auto-invoked to |
| 5 | orient a session and replaces a heavyweight master/orchestrator agent. |
| 6 | |
| 7 | **This skill produces NO planning documents.** It reads state and recommends. All |
| 8 | artifacts are produced by the specialized planning skills it routes to. |
| 9 | |
| 10 | ## When to use |
| 11 | |
| 12 | - User asks "what's next", "where am I", "what's my status", "continue", "resume". |
| 13 | - Start of any BMAD session, to orient before doing planning work. |
| 14 | - User is unsure which skill to invoke. |
| 15 | |
| 16 | ## Phase Map |
| 17 | |
| 18 | ``` |
| 19 | Analysis ──► Planning ──► Solutioning ──► Implementation-handoff |
| 20 | (optional) (required) (conditional) (required: ready-for-dev stories) |
| 21 | ``` |
| 22 | |
| 23 | | Phase | Artifact(s) | Skill | |
| 24 | |-------|-------------|-------| |
| 25 | | Analysis (optional) | product-brief, research, brainstorm | `bmad-product-brief` / `bmad-research` / `bmad-brainstorm` | |
| 26 | | Planning (required) | prd.md OR tech-spec.md, epics.md | `bmad-prd` / `bmad-tech-spec` / `bmad-epics-and-stories` | |
| 27 | | Planning — UX (only if UI) | ux-design.md | `bmad-ux` | |
| 28 | | Solutioning (track-dependent) | architecture.md | `bmad-architecture` | |
| 29 | | Implementation-handoff (required) | {epic}.{story}.{slug}.story.md (ready-for-dev) | `bmad-epics-and-stories` → `bmad-sprint-planning` | |
| 30 | |
| 31 | Threaded across every phase: **project-context.md** (the project "constitution") and |
| 32 | **decision-log.md** (decisions carried between workflows). If these are missing, the |
| 33 | first recommendation is always to establish them. |
| 34 | |
| 35 | > Implementation itself (writing code, running tests) is OUT OF SCOPE. The last artifact |
| 36 | > this plugin produces is a ready-for-dev story file or handoff manifest, passed to an |
| 37 | > external dev tool. |
| 38 | |
| 39 | ## Tracks (scale-adaptive — never numbered levels) |
| 40 | |
| 41 | A heuristic may suggest a track from story count; **the user confirms**. |
| 42 | |
| 43 | | Track | Scope | Required planning | Optional | |
| 44 | |-------|-------|-------------------|----------| |
| 45 | | **Quick Flow** | 1–15 stories | tech-spec + stories | analysis, ux | |
| 46 | | **BMad Method** | 10–50+ stories | prd + architecture + epics + stories | analysis; ux (if UI) | |
| 47 | | **Enterprise** | 30+ stories | prd + architecture + security + devops + epics + stories | analysis | |
| 48 | |
| 49 | Per-track required vs optional steps in detail: see |
| 50 | [REFERENCE.md](${CLAUDE_PLUGIN_ROOT}/skills/bmad-help/REFERENCE.md). |
| 51 | |
| 52 | ## How it works |
| 53 | |
| 54 | 1. **Detect state.** Run the detector to list which artifacts exist and the inferred |
| 55 | phase: |
| 56 | ```bash |
| 57 | bash ${CLAUDE_PLUGIN_ROOT}/skills/bmad-help/scripts/detect-state.sh [output-folder] |
| 58 | ``` |
| 59 | Default output folder is `bmad-output/`. It prints a checklist of artifacts (present |
| 60 | or missing), the detected track (from decision-log if recorded), and the inferred |
| 61 | phase. |
| 62 | |
| 63 | 2. **Recommend next.** Map the detected state to the next skill: |
| 64 | ```bash |
| 65 | bash ${CLAUDE_PLUGIN_ROOT}/skills/bmad-help/scripts/recommend-next.sh [output-folder] |
| 66 | ``` |
| 67 | It prints the single next skill to run, plus why, plus any skipped optional phases. |
| 68 | |
| 69 | 3. **Present** the recommendation to the user in plain prose. Offer to invoke the |
| 70 | recommended skill. Do not auto-run planning skills without confirmation when the |
| 71 | track is still ambiguous. |
| 72 | |
| 73 | If `project-context.md` does not exist yet, treat the project as un-initialized: the |
| 74 | recommendation is to start with `bmad-init` (or `bmad-product-brief` if the user wants the |
| 75 | optional Analysis phase first) to establish project context and choose a track. |
| 76 | |
| 77 | ## Routing logic (summary) |
| 78 | |
| 79 | Evaluate in order; recommend the first unmet step for the active track. |
| 80 | |
| 81 | 1. No `project-context.md` → establish project context (run `bmad-init`, or |
| 82 | `bmad-product-brief` if the user opts into Analysis). Confirm the track. |
| 83 | 2. No `prd.md` and no `tech-spec.md` → |
| 84 | - Quick Flow → `bmad-tech-spec` (tech-spec). |
| 85 | - BMad Method / Enterprise → `bmad-prd` (PRD). |
| 86 | 3. Project has a UI and no `ux-design.md` (BMad/Enterprise) → `bmad-ux`. |
| 87 | 4. Track needs architecture and no `architecture.md` (BMad/Enterprise) → |
| 88 | `bmad-architecture`. (Quic |