Skip to content

feat: add industry-landscape-research skill (8-step methodology)#336

Open
AaronAust1n wants to merge 1 commit into
OpenBMB:mainfrom
AaronAust1n:feat/industry-landscape-research
Open

feat: add industry-landscape-research skill (8-step methodology)#336
AaronAust1n wants to merge 1 commit into
OpenBMB:mainfrom
AaronAust1n:feat/industry-landscape-research

Conversation

@AaronAust1n

Copy link
Copy Markdown

PR: Add Industry Landscape Research Skill

Summary

This PR adds the industry-landscape-research skill — a comprehensive 8-step industry research methodology that enables PilotDeck agents to conduct professional-grade market research, competitive landscape analysis, and investment thesis development.

What This Skill Provides

A complete research methodology covering the full lifecycle of an industry analysis project:

Step Name Key Question Deliverable
1 Charter What are we answering? Research Charter (scope/exclusions/hypotheses/success criteria)
2 Decompose How to segment this industry? Track tree (2-5 main tracks + sub-ecosystems)
3 Three-Round Search Who's in the field? Company list + profiles + gap table
4 Coverage Audit Is coverage sufficient? Audit matrix (⭐ rating + 🔴/🟡 gap scoring)
5 Analysis Frameworks What does the data mean? Porter 5 Forces + 7 Powers + Wardley Map + JTBD
6 Quant Modeling Does data support the thesis? Unit economics, TAM/SAM/SOM, profit pool, sensitivity
7 Thesis Synthesis What's my view? House view + 3 scenarios + Pre-mortem + Devil's Advocate
8 Deliverable Assembly How to deliver? 6+2 chapter standard report + maps + timelines + deep-dives

Files Included (87 files, ~9,890 lines)

Core

  • SKILL.md — Main skill definition (456 lines, v2.2.0)
  • INDEX.md — Quick navigation index
  • README.md / README.zh.md — Bilingual documentation

Workflows (8 SOPs)

  • workflows/00-execution-plan.md — Phase 1/2/3 + parallel agent scheduling
  • workflows/01-research-charter.md — Scope lock with sign-off gate
  • workflows/02-three-round-search.md — R1-R8 iteration with convergence criteria
  • workflows/03-analysis-frameworks.md — Porter 5F / 7 Powers / Wardley / JTBD
  • workflows/04-coverage-audit.md — Mandatory gate before Step 5
  • workflows/05-quant-modeling.md — Unit economics, TAM, profit pool
  • workflows/06-thesis-synthesis.md — House view, scenarios, Pre-mortem
  • workflows/07-deliverable-assembly.md — 6+2 chapter standard report

Industry Archetypes (7 + 1 modifier)

  • archetypes/platform.md — Two-sided platforms
  • archetypes/saas-vertical.md — Vertical SaaS
  • archetypes/deeptech-hardware.md — DeepTech / Hardware
  • archetypes/marketplace.md — Marketplaces
  • archetypes/consumer.md — Consumer apps
  • archetypes/infrastructure.md — Infrastructure / Cloud
  • archetypes/content-media.md — Content / Media
  • archetypes/_modifier-rapidly-evolving.md — 🆕 Rapidly-evolving catalog modifier (v2.2)

Reference Guides (14 documents)

  • Source recipes for 5 deep-water industries (medical, semiconductor, auto, fintech, energy)
  • MaaS & models catalog recipe (26 sources, 9 price traps)
  • Valuation playbook, bias checklist, glossary (40+ terms), rules of thumb
  • Crawl strategy, dirty work playbook, refresh cadence, recency guardrail
  • Side-channel intel (JD, patents, investments, GitHub, app stores, Wayback)

Templates (5 sub-agent + 12 visualization + calculators)

  • 5 sub-agent templates (track scan, company deep-dive, cross-validate, devil's advocate, quant model)
  • 12 SVG/HTML visualization skeletons (landscape grid, evolution timeline, Porter radar, Wardley map, value chain, profit pool, stakeholder stance, company card, valuation leaderboard, TAM funnel, SKU pricing matrix, version timeline)
  • Unit economics calculator (Python + Excel + formulas)

Schemas (6 JSON schemas)

  • Company, event, source, racetrack, deliverable, SKU — all with validation

Examples (5 complete projects)

  • AIGC Image, LegalTech SaaS, Cross-border Marketplace, L4 Autonomous Driving, MaaS LLM Catalog

Scripts

  • validate.py — JSON Schema validator with datum quality checks
  • Makefilemake validate, make unit-economics, make all

Design Highlights

  1. Data quality enforcement: Every non-trivial statement must carry "number + source URL + fetch date + data grade (L1-L4)" — all four mandatory
  2. Coverage audit gate: Step 4 is a hard gate — audit must pass (🔴 gaps resolved) before proceeding to analysis
  3. Training cutoff mitigation (v2.2): 6 recency guardrails for rapidly-evolving catalogs (MaaS, AI tools), preventing stale model/SKU/price outputs
  4. Single Source of Truth: All deliverables derive from 3-4 data tables (companies.jsonl, events.jsonl, sources.csv, skus.jsonl)
  5. Archetype system: 7 industry archetypes + 1 orthogonal modifier — compose the right analysis framework per industry
  6. Parallel agent orchestration: Templates for slicing by track, company, data source, geography, or opposing viewpoint

Checklist

  • SKILL.md with valid YAML frontmatter (name + description + version)
  • All supporting files are self-contained (no external dependencies beyond Python stdlib)
  • Bilingual documentation (README.md + README.zh.md)
  • JSON schemas validate correctly with scripts/validate.py
  • Example projects include real sample data
  • Compatible with Anthropic SKILL.md format

Add comprehensive 8-step industry research methodology skill for
PilotDeck. Covers: research charter, track decomposition, 3-round
search + multi-iteration, company profiling, valuation tracking,
coverage audit, analysis frameworks (Porter 5 Forces / 7 Powers /
Wardley / JTBD), unit economics, thesis synthesis, and deliverable
assembly.

Includes:
- 8 workflow SOPs (00-07)
- 7 industry archetypes + 1 rapidly-evolving modifier
- 14 reference guides (source recipes, valuation, bias, etc.)
- 12 visualization templates (SVG/HTML)
- 5 sub-agent templates
- 6 JSON schemas for data validation
- 5 example projects with sample data
- Python validator + Makefile
- Unit economics calculator

87 files total. Skill format: Anthropic SKILL.md + YAML frontmatter.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant