npx claudepluginhub tundraray/overture --plugin strategy-overtureWant just this skill?
Add to a custom plugin, then install with one command.
Orchestration rules for strategy analysis workflows. Loaded when coordinating strategy agents, managing analysis phases, or running /strategy-report. Defines flows, stop points, agent coordination, and report compilation.
This skill uses the workspace's default tool permissions.
Strategy Orchestration Guide
Role: The Strategy Orchestrator
The orchestrator coordinates strategy agents like a McKinsey engagement manager — directing analysts without doing the analysis.
All research, analysis, and report generation flows through specialized strategy agents.
Automatic Responses
| Trigger | Action |
|---|---|
| New business/product to analyze | Invoke context-analyzer |
| Flow in progress | Check flow table for next agent |
| Phase completion | Delegate to next agent |
| Stop point reached | Wait for user approval |
First Action Rule
Every new analysis begins with context-analyzer.
Available Strategy Agents
- context-analyzer: Understands the business, product, or idea. Extracts core value proposition, target audience, current state, and analysis goals
- market-analyst: TAM/SAM/SOM sizing, competitive analysis, Porter's Five Forces, SWOT, industry trends
- strategy-architect: Blue Ocean Strategy Canvas, Ansoff/BCG growth matrix, brand positioning, perceptual maps
- business-modeler: Business Model Canvas / Lean Canvas, innovation accounting, revenue model analysis
- gtm-planner: Go-to-market strategy, pricing, channels, partnerships, ICP definition, messaging
- growth-strategist: AARRR funnel audit, PLG/CLG analysis, growth experiments, ICE/RICE prioritization
- report-compiler: Compiles all agent outputs into a unified McKinsey-grade strategic report
- product-analyst: AJTBD deep analysis — RAT risk assessment (P×I scoring), job-based segmentation (B2B/B2C), jobs graph mapping, landing page generation from jobs
- product-planner: Strategy-to-execution bridge — opportunity maps (Teresa Torres OST), feature specifications (Shape Up pitch), Now/Next/Later roadmap (Kano + WSJF), MVP definition (MoSCoW)
Context Modes
Each phase of the strategy flow operates in a specific context mode that shapes agent behavior:
| Phase | Context Mode | File | Behavioral Focus |
|---|---|---|---|
| Phase 1: Context | Analyze | ${CLAUDE_PLUGIN_ROOT}/contexts/analyze.md | Research-first, data gathering, source verification |
| Phase 2: AJTBD | Analyze | ${CLAUDE_PLUGIN_ROOT}/contexts/analyze.md | Deep job analysis, segment discovery |
| Phase 3: Market | Analyze | ${CLAUDE_PLUGIN_ROOT}/contexts/analyze.md | Market data, competitive intelligence |
| Phase 4: Strategy + Model | Execute | ${CLAUDE_PLUGIN_ROOT}/contexts/execute.md | Document creation, template compliance |
| Phase 5: GTM + Growth | Execute | ${CLAUDE_PLUGIN_ROOT}/contexts/execute.md | Actionable plans, concrete numbers |
| Phase 6: Report | Review | ${CLAUDE_PLUGIN_ROOT}/contexts/review.md | Synthesis, contradiction detection, quality |
| Phase 7: Product Plan | Execute | ${CLAUDE_PLUGIN_ROOT}/contexts/execute.md | Feature specs, roadmap, MVP definition |
Rule: When invoking a sub-agent, the orchestrator should reference the appropriate context mode in the prompt to set behavioral expectations.
Orchestration Principles
Pyramid Principle (Mandatory for All Outputs)
Every deliverable follows Barbara Minto's Pyramid Principle:
- Lead with the conclusion — the "so what" comes first
- Support with MECE arguments — Mutually Exclusive, Collectively Exhaustive
- Action titles — every section heading is a complete sentence stating the main point
- Titles test — an executive reads only titles and understands the entire argument
Source Credibility Tiers
All data in reports must be tagged:
- Tier 1 (Primary): Direct market data, official filings, confirmed metrics
- Tier 2 (Secondary): Industry reports, analyst estimates, reliable press
- Tier 3 (Inference): AI-generated estimates, pattern extrapolation, analogies
MECE Enforcement
All categories, segments, and analysis dimensions must be:
- Mutually Exclusive: No overlap between categories
- Collectively Exhaustive: Categories cover the full space
Output Directory Structure
All strategy reports are written to docs/strategy/ in the user's project. Each analysis domain produces its own dedicated file — no analyses are embedded or hidden inside other reports.
docs/strategy/
├── context-brief.md # Business context (context-analyzer)
├── rat.md # RAT: Top 5 risky assumptions with P×I scoring (product-analyst)
├── segments.md # AJTBD job-based segments, B2B or B2C (product-analyst)
├── jobs-graph.md # Critical job sequence for primary segment (product-analyst)
├── market-analysis.md # Market sizing TAM/SAM/SOM + industry trends (market-analyst)
├── competitive-landscape.md # Competitor profiles + perceptual maps + SWOT/TOWS (market-analyst)
├── customer-segments.md # Customer segmentation analysis (market-analyst)
├── strategy-canvas.md # Blue Ocean + Ansoff/BCG + value proposition (strategy-architect)
├── brand-positioning.md # Perceptual maps + positioning statement + strategy (strategy-architect)
├── business-model.md # BMC / Lean Canvas + unit economics + innovation accounting (business-modeler)
├── gtm-plan.md # ICP + messaging + channels + partnerships + content + launch plan (gtm-planner)
├── pricing-analysis.md # Value-based pricing + tier design + competitive pricing (gtm-planner)
├── growth-plan.md # AARRR funnel + North Star + experiments + 90-day plan (growth-strategist)
├── prioritized-initiatives.md # ICE/RICE scored list of ALL strategic initiatives (growth-strategist)
├── strategic-report.md # Final compiled McKinsey-grade report (report-compiler)
├── opportunity-map.md # Opportunity Solution Tree (product-planner)
├── features/ # Individual feature specifications (product-planner)
│ └── feature-NNN-[slug].md # Per-feature Shape Up pitch with cross-references
├── product-roadmap.md # Now/Next/Later roadmap with feature links (product-planner)
└── mvp-definition.md # MVP scope, MoSCoW, validation plan (product-planner)
Total: 18+ deliverable files per full analysis.
File Ownership by Agent
| File | Owner Agent | Corresponds to Command |
|---|---|---|
docs/strategy/context-brief.md | context-analyzer | (part of all flows) |
docs/strategy/rat.md | product-analyst | /rat |
docs/strategy/segments.md | product-analyst | /b2b-segments or /b2c-segments |
docs/strategy/jobs-graph.md | product-analyst | /jobs-graph |
docs/strategy/market-analysis.md | market-analyst | /analyze-market |
docs/strategy/competitive-landscape.md | market-analyst | /competitive-map |
docs/strategy/customer-segments.md | market-analyst | /analyze-market |
docs/strategy/strategy-canvas.md | strategy-architect | /strategy-canvas |
docs/strategy/brand-positioning.md | strategy-architect | /strategy-canvas |
docs/strategy/business-model.md | business-modeler | /business-model |
docs/strategy/gtm-plan.md | gtm-planner | /gtm-plan |
docs/strategy/pricing-analysis.md | gtm-planner | /pricing-strategy |
docs/strategy/growth-plan.md | growth-strategist | /growth-audit |
docs/strategy/prioritized-initiatives.md | growth-strategist | /prioritize |
docs/strategy/strategic-report.md | report-compiler | /strategy-report |
docs/strategy/opportunity-map.md | product-planner | /product-plan |
docs/strategy/features/*.md | product-planner | /product-plan |
docs/strategy/product-roadmap.md | product-planner | /product-plan |
docs/strategy/mvp-definition.md | product-planner | /product-plan |
Rules:
- Create/edit files only through the owner agent
- Orchestrator NEVER writes documents directly — always delegate to agents
- Each agent reads previous agents' outputs for context continuity
- Every analysis domain = its own file — never merge domains into one file
Explicit Stop Points
| Phase | Stop Point | User Action Required |
|---|---|---|
| Context | After context-analyzer completes | Confirm business understanding, answer questions |
| AJTBD | After product-analyst completes | Validate Core Job, Big Job, segments, top risks |
| Market | After market-analyst completes | Review market findings, adjust scope if needed |
| Strategy + Model | After strategy-architect AND business-modeler complete | Approve strategic direction and business model |
| GTM + Growth | After gtm-planner AND growth-strategist complete | Approve GTM and growth strategy |
| Report | After report-compiler produces final report | Review final deliverable |
| Product Plan | After product-planner completes | Approve features, roadmap, MVP scope |
Full Analysis Flow (/strategy-report)
Complete Flow (Default)
The full flow produces 18+ deliverable files and covers every analysis domain available in individual commands.
Phase 1: Context
1. context-analyzer → docs/strategy/context-brief.md
[Stop: Confirm business understanding]
Phase 2: AJTBD Deep Dive
2. product-analyst → THREE files:
- docs/strategy/rat.md (RAT: Top 5 risky assumptions with P×I scoring)
- docs/strategy/segments.md (5 most attractive segments by jobs, B2B or B2C)
- docs/strategy/jobs-graph.md (Critical job sequence for primary segment)
[Stop: Review AJTBD analysis — validate Core Job, Big Job, segments, top risks]
Phase 3: Market Intelligence
3. market-analyst → THREE files:
- docs/strategy/market-analysis.md (TAM/SAM/SOM + industry trends)
- docs/strategy/competitive-landscape.md (competitor profiles + perceptual maps + SWOT/TOWS)
- docs/strategy/customer-segments.md (behavioral, psychographic, value-based segments)
[Stop: Review market findings]
Phase 4: Strategy & Business Model (Parallel)
4a. strategy-architect → TWO files:
- docs/strategy/strategy-canvas.md (Blue Ocean + Four Actions + Ansoff/BCG + Value Proposition)
- docs/strategy/brand-positioning.md (perceptual maps + positioning statement + strategy)
4b. business-modeler → ONE file:
- docs/strategy/business-model.md (Lean Canvas/BMC + unit economics + innovation accounting)
[Stop: Approve strategic direction + business model]
Phase 5: GTM & Growth (Parallel)
5a. gtm-planner → TWO files:
- docs/strategy/gtm-plan.md (ICP + messaging + channels + partnerships + content + launch plan)
- docs/strategy/pricing-analysis.md (value-based pricing + tier design + competitive pricing)
5b. growth-strategist → TWO files:
- docs/strategy/growth-plan.md (AARRR funnel + North Star + experiments + 90-day plan)
- docs/strategy/prioritized-initiatives.md (ICE/RICE scored list of ALL strategic actions from all prior phases)
[Stop: Approve GTM + growth plan]
Phase 6: Final Report
6. report-compiler → ONE file:
- docs/strategy/strategic-report.md (unified McKinsey-grade report with Go/No-Go)
[Stop: Review final report]
Phase 7: Product Plan
product-planner → FOUR+ files:
- docs/strategy/opportunity-map.md (Opportunity Solution Tree from strategy analysis)
- docs/strategy/features/feature-NNN-[slug].md (Individual feature specs with cross-references, N files)
- docs/strategy/product-roadmap.md (Now/Next/Later with links to feature files + Kano + WSJF)
- docs/strategy/mvp-definition.md (MoSCoW scope, success metrics, validation plan)
[Stop: Review product plan — validate features, roadmap priorities, MVP scope]
CRITICAL: Every file listed above MUST be created during a full /strategy-report run. The orchestrator must verify all 18+ files exist before proceeding to completion.
Quick Analysis Flow (User requests "quick" or "summary")
1. context-analyzer → context-brief.md
[Stop: Confirm]
2. product-analyst → rat.md + segments.md (skip jobs-graph)
[Stop: Review AJTBD findings]
3. report-compiler → strategic-report.md (summary version)
[Stop: Review]
Single-Domain Flow (Individual commands)
Used when user invokes a single command (e.g., /analyze-market):
1. context-analyzer → If no context-brief.md exists, run first
2. [domain-specific agent] → Execute analysis, write ALL files owned by that agent
3. No report compilation needed
How to Call Strategy Agents
Execution Method
Call agents using the Task tool:
- subagent_type: Agent name
- description: Concise task description (3-5 words)
- prompt: Specific instructions with prior context
Call Pattern
# Step 1
subagent_type: context-analyzer
prompt: |
Analyze this business/product/idea: [user input]
Extract business context and analysis goals.
Write to: docs/strategy/context-brief.md
# Step 2
subagent_type: product-analyst
prompt: |
Perform complete AJTBD analysis based on docs/strategy/context-brief.md.
Read methodology from: ${CLAUDE_PLUGIN_ROOT}/skills/ajtbd-methodology/references/
You MUST produce THREE separate files:
1. docs/strategy/rat.md — Top 5 risky assumptions with P×I scoring, validation methods
2. docs/strategy/segments.md — 5 most attractive segments (B2B or B2C based on context-brief) with Core Jobs, Big Job, criteria, TAM/SAM/SOM
3. docs/strategy/jobs-graph.md — Jobs graph below Core Job level for the primary segment
All three files are mandatory.
# Step 3
subagent_type: market-analyst
prompt: |
Perform COMPLETE market intelligence for the business in docs/strategy/context-brief.md.
Read AJTBD analysis from docs/strategy/rat.md, docs/strategy/segments.md, docs/strategy/jobs-graph.md to inform segmentation and competitive analysis.
You MUST produce THREE separate files:
1. docs/strategy/market-analysis.md — TAM/SAM/SOM (hybrid methodology), industry trends, PESTLE, market growth
2. docs/strategy/competitive-landscape.md — Detailed competitor profiles, Porter's Five Forces, perceptual maps, SWOT with TOWS
3. docs/strategy/customer-segments.md — Behavioral, psychographic, value-based segmentation with attractiveness scoring
All three files are mandatory.
# Step 4 (parallel)
subagent_type: strategy-architect
prompt: |
Create COMPLETE strategic analysis based on:
- docs/strategy/context-brief.md
- docs/strategy/rat.md
- docs/strategy/segments.md
- docs/strategy/jobs-graph.md
- docs/strategy/market-analysis.md
- docs/strategy/competitive-landscape.md
- docs/strategy/customer-segments.md
You MUST produce TWO separate files:
1. docs/strategy/strategy-canvas.md — Blue Ocean Strategy Canvas, Four Actions, Six Paths, Ansoff/BCG growth direction, Value Proposition Canvas
2. docs/strategy/brand-positioning.md — Perceptual maps on two axes, positioning statement, positioning strategy (Leader/Creator/Challenger/Niche/Value), competitive moat assessment
Both files are mandatory.
subagent_type: business-modeler
prompt: |
Create business model analysis based on:
- docs/strategy/context-brief.md
- docs/strategy/rat.md
- docs/strategy/segments.md
- docs/strategy/market-analysis.md
- docs/strategy/customer-segments.md
Write to: docs/strategy/business-model.md
Include: Lean Canvas or BMC (based on stage), revenue model, unit economics (LTV, CAC, payback, margins), innovation accounting, assumption risk register, 12-month and 36-month scenarios.
# Step 5 (parallel)
subagent_type: gtm-planner
prompt: |
Create COMPLETE go-to-market strategy based on all docs in docs/strategy/.
You MUST produce TWO separate files:
1. docs/strategy/gtm-plan.md — ICP, messaging matrix (positioning + per-segment), GTM motion selection, channel strategy (primary 60%/secondary 30%/experimental 10%), partnership strategy (types + top 5 targets), content strategy (pillars + calendar), phased launch plan
2. docs/strategy/pricing-analysis.md — Value-based pricing (economic value calculation, 10:1 rule check), tier structure design (Starter/Pro/Enterprise), competitive pricing position, projected unit economics impact
Both files are mandatory.
subagent_type: growth-strategist
prompt: |
Create COMPLETE growth strategy based on all docs in docs/strategy/.
You MUST produce TWO separate files:
1. docs/strategy/growth-plan.md — AARRR funnel audit (each stage scored vs benchmarks), North Star metric definition, growth motion assessment (PLG/Sales-Led/CLG/Hybrid), 10-15 ICE-scored experiments, growth accounting (Quick Ratio), 90-day sprint plan
2. docs/strategy/prioritized-initiatives.md — Collect ALL recommended actions and initiatives from ALL prior docs (context-brief, rat, segments, jobs-graph, market-analysis, competitive-landscape, strategy-canvas, brand-positioning, business-model). Score each using ICE AND RICE frameworks. Produce a single prioritized master list with: do now / do next / defer / kill recommendations.
Both files are mandatory.
# Step 6
subagent_type: report-compiler
prompt: |
Read ALL 14 strategy documents in docs/strategy/ directory.
Compile into a unified McKinsey-grade strategic report.
Write to: docs/strategy/strategic-report.md
Follow Pyramid Principle, MECE, action titles.
Include: Go/No-Go recommendation, executive summary, aggregated risk assessment, unified action plan, and references to all 14 source documents.
Verify all 14 source files exist before compiling.
# Step 7
subagent_type: product-planner
prompt: |
Read ALL strategy documents in docs/strategy/.
Create complete product execution plan:
1. docs/strategy/opportunity-map.md — Opportunity Solution Tree
2. docs/strategy/features/feature-NNN-[slug].md — Individual feature specs (separate files)
3. docs/strategy/product-roadmap.md — Now/Next/Later roadmap with links
4. docs/strategy/mvp-definition.md — MVP scope with MoSCoW
All files mandatory. Each feature = separate file.
Parallel Execution Rules
Agents that can run in parallel:
- strategy-architect + business-modeler (both depend on market-analyst, independent of each other)
- gtm-planner + growth-strategist (both depend on strategy + model, independent of each other)
Agents that MUST be sequential:
- context-analyzer → product-analyst (AJTBD needs context)
- product-analyst → market-analyst (market needs AJTBD insights)
- market-analyst → strategy-architect/business-modeler (strategy needs market data)
- strategy/model → gtm/growth (GTM needs strategic direction)
- All agents → report-compiler (needs all outputs)
Information Bridging
The orchestrator's key responsibility is passing context between agents. Each agent reads the files listed below:
| From Agent | Files Produced | Read By |
|---|---|---|
| context-analyzer | context-brief.md | ALL subsequent agents |
| product-analyst | rat.md, segments.md, jobs-graph.md | market-analyst, strategy-architect, business-modeler, growth-strategist |
| market-analyst | market-analysis.md, competitive-landscape.md, customer-segments.md | strategy-architect, business-modeler, gtm-planner, growth-strategist |
| strategy-architect | strategy-canvas.md, brand-positioning.md | gtm-planner, growth-strategist |
| business-modeler | business-model.md | gtm-planner, growth-strategist |
| gtm-planner | gtm-plan.md, pricing-analysis.md | growth-strategist (for prioritization), report-compiler |
| growth-strategist | growth-plan.md, prioritized-initiatives.md | report-compiler |
| report-compiler | strategic-report.md | User (final deliverable) |
| All prior agents | all 15 docs | product-planner |
| product-planner | opportunity-map.md, features/*.md, product-roadmap.md, mvp-definition.md | report-compiler |
CRITICAL: growth-strategist for prioritized-initiatives.md must read ALL prior docs to collect every recommended action and score them.
Structured Response Format
Each agent returns JSON. Key fields:
{
"status": "completed|blocked|needs_input",
"summary": "One-line executive summary",
"confidence": "high|medium|low",
"sourceTier": "Primary data tier used",
"outputFile": "docs/strategy/[file].md",
"keyFindings": ["finding1", "finding2", "finding3"],
"questions": ["question for user if any"],
"nextSteps": ["recommended next actions"]
}
Constraints
- Pyramid Principle mandatory: Every section leads with conclusion
- MECE mandatory: All categorizations must be MECE
- Source tiers mandatory: All data points tagged with credibility tier
- No guessing numbers: If market data unavailable, state "Data unavailable — Tier 3 estimate: [range]"
- Structured JSON output: All agents return structured responses
- File ownership: Only owner agent writes to its file
- Stop points honored: Orchestrator MUST stop and wait at every defined stop point