[Documentation] Use when generating, smart-merging or refactoring the root AI context (CLAUDE.md, AGENTS.md) and its Claude/Codex projections. Config file only: project-config.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
[Research] Use when a workflow step or the user asks for market analysis: competitors, TAM/SAM/SOM, trends, SWOT and customer segments.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
설치 방법 보기설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.
Goal: Analyze a market landscape and deliver an evidence-backed market-analysis artifact—competitors, TAM/SAM/SOM, trends, SWOT, and customer segments—that downstream skills can consume as evidence.
Summary:
business-evaluation and strategy-builder, not unsupported opinion.web-research --chain=deep-dive (_evidence-{slug}.md + _sources-{slug}.md), read it before any search and cite it; search only its named gaps, at most 5 searches. Every market figure (TAM/SAM/SOM, growth, share) rests on 2+ independent sources.[N] mapped to Sources or N/A — {reason}; every size claim cites source/year/methodology; never invent source metadata; SWOT requires evidence.workflow-research business-eval/marketing uses parent ARTIFACT_SLUG + MARKET_ANALYSIS_PATH exactly; otherwise use the fallback path; return the exact written path and copy the plan-dir path when required.Workflow:
Key Rules:
WebSearch calls and 5 WebFetch calls. With no evidence base, stay within the web-research budget (at most 10 searches, then at most 8 fetches)single source, unverified and hold it below 60% confidence[N] citation that maps to one Sources row; use N/A — {reason} when evidence is unavailableBe skeptical. Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence percentages (Idea should be more than 80%).
Before research, classify scope and available evidence as present, missing, or ambiguous; clarify missing or ambiguous industry, geography, timeframe, and focus with user.
Load the research evidence base first. When web-research --chain=deep-dive (or web-research followed by source-deep-dive) ran for this topic, read its evidence base tmp/research/_evidence-{slug}.md and source map tmp/research/_sources-{slug}.md before any search. Use the exact paths that research step returned, or a path the user supplies; never guess a slug. Build Steps 2-6 from it and carry its sources into the Sources table with their Title, URL, Author/Publisher, Date and Tier; do not re-search what it already establishes. List what Steps 2-6 still need that it lacks (its ## Gaps Remaining, a missing competitor, a sizing report) and search only those named gaps: at most 5 WebSearch calls and 5 WebFetch calls. With no evidence base, research under the web-research budget: at most 10 searches, then at most 8 fetches.
Cross-validate every market figure. Each TAM/SAM/SOM value, growth rate and market share rests on 2+ independent sources, the same rule the research step applies; sources that copy one original study count once. A single-source figure is marked single source, unverified and held below 60% confidence; conflicting figures are shown side by side with the one you weight and why.
Analysis lens: For each step, run one focused Think: pass: what evidence supports, limits, or falsifies each finding?
Confirm scope with user:
For each competitor (identify 5-10):
| Field | Source |
|---|---|
| Company name | Evidence base, then WebSearch for gaps |
| Positioning/tagline | Company website |
| Key products | Product pages |
| Pricing model | Pricing page or reports |
| Strengths | Reviews, analyst reports |
| Weaknesses | Reviews, customer complaints |
| Market share (if available) | Industry reports |
Use Tier 1-2 sources (Gartner, Statista, IBISWorld, government data):
Every number must cite source, year, methodology.
Cross-validate each figure against a second independent source (see the precondition); a figure with one source stays single source, unverified, below 60% confidence.
Research and categorize:
Each item must link to evidence:
| Category | Item | Evidence |
|---|---|---|
| Strength | {item} | Source [N] |
| Weakness | {item} | Source [N] |
| Opportunity | {item} | Source [N] |
| Threat | {item} | Source [N] |
For each segment:
MANDATORY IMPORTANT MUST ATTENTION Write the result via enforced .claude/templates/market-analysis-template.md to parent-provided MARKET_ANALYSIS_PATH when present; otherwise use docs/knowledge/strategy/market-analysis/{descriptive-slug}.md — why: downstream skills consume this as EVIDENCE; an informal or inline handoff without deterministic location and known shape fails silently, leaving consumers nothing to cite.
When invoked by workflow-research in business-eval or marketing mode, use parent-provided
ARTIFACT_SLUG and MARKET_ANALYSIS_PATH exactly; do not derive a new slug. Return the exact written
path in the completion handoff so the next skill can verify it before reading.
Consumed by:
business-evaluation skill (business viability) — reads Sizing, Competitors, Trends, SWOT, Segmentsstrategy-builder skill (marketing strategy)When invoked inside a workflow that also writes to a plan directory, copy the file to {plan-dir}/research/market-analysis.md so the plan artifact set stays self-contained.
[IMPORTANT] Use
TaskCreateto break ALL work into small tasks BEFORE starting.
IMPORTANT MUST ATTENTION Goal: Analyze a market landscape and deliver an evidence-backed market-analysis artifact—competitors, TAM/SAM/SOM, trends, SWOT, and customer segments—that downstream skills can consume as evidence.
IMPORTANT MUST ATTENTION Main path: Classify and clarify scope/evidence; then run in order: (1) research 5-10 competitors, (2) size TAM/SAM/SOM, (3) analyze growth, disruption, regulation, and behavior trends, (4) evidence-link SWOT, and (5) segment customers by demographics, psychographics, behavior, and jobs-to-be-done. Write via the enforced template and complete the exact handoff path. NEVER skip, reorder, or merge steps.
IMPORTANT MUST ATTENTION Modes/gates: In workflow-research business-eval/marketing, use parent ARTIFACT_SLUG + MARKET_ANALYSIS_PATH exactly; otherwise use the fallback path; return the exact written path; copy the plan-dir artifact when required. Prefer Tier 1-2 sizing sources; every factual claim, number, table row, and inference ends [N] mapped to Sources or N/A — {reason}; cite size source/year/methodology; never invent metadata; link every SWOT item to evidence.
IMPORTANT MUST ATTENTION Evidence base: read this run's _evidence-{slug}.md + _sources-{slug}.md first when they exist and cite them; search only named gaps (at most 5 searches, 5 fetches; no evidence base → at most 10 searches, 8 fetches); every TAM/SAM/SOM, growth and share figure needs 2+ independent sources, else single source, unverified below 60%.
IMPORTANT MUST ATTENTION Apply critical/sequential thinking; run one focused Think: pass per step; cite source evidence and confidence; preserve uncertainty; never present unsupported findings as fact.
IMPORTANT MUST ATTENTION Create TaskCreate todos before starting; keep one in_progress, mark each completed with evidence, and add a final review todo. If task tools are unavailable, maintain equivalent synchronized statuses.
Anti-Rationalization:
| Evasion | Rebuttal |
|---|---|
| "Market size is obvious; source later." | Cite source, year, and methodology now; use N/A — {reason} when unavailable. |
| "One competitor or source is enough." | Research 5-10 competitors; prefer Tier 1-2 sources and preserve metadata. |
| "A fresh search is faster than reading the evidence base." | Read the evidence base first and cite it; search only its named gaps, within the cap. |
| "One report gives the market size." | Cross-validate with a second independent source, or mark the figure single source, unverified below 60%. |
| "SWOT is just judgment." | Link every item to evidence; label unavailable support N/A — {reason}. |
| "Output path is optional." | Use the enforced template/path contract, return the exact written path, and copy the plan-dir artifact when required. |
IMPORTANT MUST ATTENTION verify all claims, numbers, source metadata, and handoff paths before completion; NEVER guess, skip evidence, or weaken the output contract.
아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.
로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.
아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.
[Documentation] Use when generating, smart-merging or refactoring the root AI context (CLAUDE.md, AGENTS.md) and its Claude/Codex projections. Config file only: project-config.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
[Code Quality] Use when a workflow step or the user asks for AI code or plan review: LLM calls, prompts, agents, RAG, evals and guardrails. --mode={code|plan}; --report-only is read-only.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
[Architecture] Use when a workflow step or the user asks for architecture design, compliance review, scalability grading or whole-project audit via --mode={design|review|scalability|full}.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
[Content] Use when a workflow step or the user asks for product/feature ideation with ranked opportunities and validation tests. --mode={roadmap|scope}.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
[Content] Use when a workflow step or the user asks for business viability: Business Model Canvas, financials, risks and go-to-market recommendation.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
[Code Quality] Use when a workflow step or the user asks for a focused code-change review: staged, unstaged, branch diff or simple pull request. --fix-loop fixes and re-reviews. Complex pull request: workflow-review-changes; rationale: why-review.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.