create-dss-project 0.1.0

package/template/CLAUDE.md

@@ -0,0 +1,145 @@
+# CLAUDE.md
+
+## Current Date
+
+Always use `date` via Bash to get today's date at the start of a session.
+
+## What We're Building
+
+**{{PROJECT_NAME}}** — {{ONE_LINE_DESCRIPTION}}. Goal: {{GOAL}}. Team: {{TEAM}}.
+
+## Scope
+
+{{SCOPE_DEFINITION}}. Out of scope: {{OUT_OF_SCOPE}}.
+
+## Core Bet
+
+{{STRATEGIC_HYPOTHESIS}}
+
+## Target Customer
+
+{{ICP_DESCRIPTION}}
+
+---
+
+## Project Configuration
+
+Project settings are stored in `project.config.json` (generated by `/onboard`). Skills read this file for project name, entity types, module flags, and scoring dimensions. See `project.config.example.json` for the schema.
+
+---
+
+## Context Loading
+
+Load the snapshot for your task type. **Do not read raw source files unless the snapshot is insufficient for your task.**
+
+| Task | Load first |
+|------|-----------|
+| Discovery call prep, hypothesis work, strategic review | `context/project-state.md` |
+| Pipeline updates, outreach, call scheduling | `context/pipeline-state.md` |
+| Competitor or market research | `context/competitor-snapshot.md` + `context/market-snapshot.md` |
+| End-of-session (`/session-end`) | `context/project-state.md` (verify what changed this session) |
+
+### Three Loading Modes
+
+| Mode | When | Rule |
+|------|------|------|
+| **Fast** | Pipeline updates, housekeeping, scheduling, call prep | Snapshot only |
+| **Standard** | Research, competitor analysis, entity enrichment | Snapshot + targeted raw files as needed |
+| **Deep** | Hypothesis review, critical reasoning, kill condition updates, strategic decisions | Load snapshot for orientation — then read ALL raw evidence before forming any view |
+
+**Deep mode rule**: Snapshots tell you what was concluded; raw files contain disconfirming signals, hedged language, and contradictions that drive good reasoning. For hypothesis-class work, reading only snapshots is not acceptable.
+
+Raw evidence sources for Deep mode:
+- All files in `discovery/calls/` — every call note
+- `memory/decisions.md` — full decision log with rationale
+- `memory/research.md` — capability map and market findings
+
+---
+
+## Multi-Agent Coordination
+
+### Before Starting
+1. Run `/session-start` — it checks conflicts, loads context, and shows priorities.
+2. Add yourself to "In Progress" with workstream, item, date.
+3. Use templates from `templates/`.
+
+### While Working
+- One file per unit of work.
+- Research → `research/`. Discovery → `discovery/`. Docs → `docs/`.
+- Never overwrite another agent's in-progress file without reading it first.
+
+### Pipeline Work
+Dashboard source of truth: `{{PIPELINE_SOURCE_OF_TRUTH}}` — add {{ENTITY_TYPE_PLURAL}} here first, never only to JSON.
+
+### After Completing Work
+1. Move STATUS.md row to "Completed Recently".
+2. Update the relevant `context/` snapshot — keep it current for the next session.
+3. If work changes hypothesis or kill conditions → run `/critical-reasoning` before updating `docs/executive-summary.md`.
+
+### Executive Summary Rule
+`docs/executive-summary.md` is the single source of truth. Update after: every discovery call, kill condition changes, major research, strategic decisions.
+
+### Agent Briefing Format
+Task → Constraints → 3–5 key facts (with file paths, not content) → output spec → model weight. No file content dumps.
+
+---
+
+## Model Routing
+
+Default: Opus. Switch to Sonnet for mechanical/template work (`/model sonnet`).
+
+Opus required: hypothesis review, discovery call processing, critical reasoning, strategic decisions.
+Sonnet fine: competitor research, call prep, housekeeping, dashboard/data work.
+
+---
+
+## End-of-Session Housekeeping (MANDATORY)
+
+Run `/session-end`. This executes the full 10-step housekeeping checklist automatically (file audit, memory cascade, hypothesis review, status sync, dashboard rebuild, commit). Reference: `docs/memos/housekeeping-reference.md`.
+
+After any session that changes discovery, pipeline, or decisions — `/session-end` will regenerate the affected `context/` snapshot via `/rebuild-snapshots`.
+
+Evidence grading rules: `docs/memos/evidence-grading.md`
+
+---
+
+## Project-Level Skills
+
+<!-- These skills live in this project's skills/ directory -->
+
+| Skill | Phase | Description |
+|-------|-------|-------------|
+| `/onboard` | Setup | One-time project configuration wizard |
+| `/session-start` | Session | Load context, check conflicts, show priorities |
+| `/session-end` | Session | 10-step end-of-session housekeeping |
+| `/health-check` | Quality | Project integrity audit and health score |
+| `/rebuild-snapshots` | Session | Regenerate all context snapshots from raw files |
+| `/pipeline-update` | Pipeline | Track entity status transitions |
+| `/outreach-sequence` | Pipeline | Design multi-touch outreach cadences |
+| `/process-call` | Discovery | Post-discovery-call structured processing |
+| `/enrich-entity` | Research | Deep research and dashboard enrichment for an entity |
+| `/synthesise` | Research | Cross-file research synthesis into structured memos |
+| `/critical-reasoning` | Analysis | Pressure-test ideas with 4 lenses: truth, consequences, risks, feasibility |
+| `/decision` | Analysis | Record strategic decisions with full rationale |
+| `/compare-options` | Analysis | Score and compare 2–5 strategic options |
+| `/weekly-report` | Reporting | Generate stakeholder-ready weekly summaries |
+
+---
+
+## Relevant User-Level Skills
+
+<!-- These skills live at ~/.claude/skills/ and are available across all projects.
+     Add project-specific context sections to them if needed. -->
+
+- `/bottom-up-tam`, `/micro-segmentation` — market sizing
+- `/critical-reasoning` — hypothesis stress-testing (project-level version is domain-agnostic)
+- `/strategic-framework-synthesiser` — structured analysis (Porter's, PESTLE, BCG, etc.)
+- `/gtm-playbook` — go-to-market planning
+- `/outside-in-benchmarking` — competitive analysis
+- `/pricing-packaging` — monetisation design
+- `/unit-economics-modeler` — CAC, LTV, payback period
+- `/burning-platform` — change management narratives
+- `/macro-trend-synthesis` — PESTLE+ trend analysis
+- `/partnership-ecosystem-strategy` — partner program design
+- `/product-roadmap-prioritisation` — backlog prioritisation
+- `/profit-pool-mapping` — value chain analysis

package/template/LICENSE

@@ -0,0 +1,21 @@
+MIT Licence
+
+Copyright (c) 2026 DS Strategy Stack Contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicence, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/template/README.md

@@ -0,0 +1,201 @@
+# DS Strategy Stack (Claude Code)
+
+> An open-source, AI-powered strategy and project management stack for Claude Code. Clone it, run `/onboard`, and start doing real work — market entry, competitor research, internal implementations, due diligence, or any corporate project that needs structured thinking.
+
+[![MIT Licence](https://img.shields.io/badge/licence-MIT-blue.svg)](LICENSE)
+
+---
+
+## What Is This?
+
+A complete strategy project framework that runs inside [Claude Code](https://claude.ai/code). It gives you:
+
+- **14 skills** that automate strategy workflows (research, discovery, analysis, reporting)
+- **A live dashboard** (Vercel-deployable) showing kill conditions, pipeline, competitors, decisions, and more
+- **A memory layer** that persists strategic context across sessions
+- **Multi-agent coordination** so multiple Claude instances can work on the same project
+- **Evidence grading** on every claim — `[CONFIRMED]`, `[SECONDARY]`, `[INFERENCE]`, `[ASSUMPTION]`
+- **Kill condition tracking** — falsifiable thresholds that tell you when to stop or pivot
+
+No boilerplate. No blank-page problem. Clone, configure, and go.
+
+![Dashboard Screenshot](dashboard/screenshot.png)
+
+---
+
+## Quick Start
+
+### Option A: npx (recommended)
+
+```bash
+npx create-dss-project my-project
+cd my-project
+claude
+/onboard
+```
+
+The scaffolder asks 3 questions (project name, type, structure level) and sets up a ready-to-go project with the right modules pre-configured.
+
+### Option B: Git clone
+
+```bash
+git clone https://github.com/DiffTheEnder/DSS-Claude-Stack.git my-project
+cd my-project
+claude
+/onboard
+```
+
+Both paths end at `/onboard`, which walks you through the full configuration — **Quick Start** (7 questions) or **Full Setup** (16 questions). It asks about your goals, experience level, and which features you want, then generates your `project.config.json`, populates placeholders, and activates the right modules.
+
+> **New to coding?** See the [Getting Started Guide](docs/getting-started.md) for step-by-step instructions covering installation, setup, and Vercel deployment — no programming knowledge needed.
+
+---
+
+## Features
+
+### Skills Library (14 skills)
+
+| Skill | Phase | Description |
+|-------|-------|-------------|
+| `/onboard` | Setup | One-time project configuration wizard |
+| `/session-start` | Session | Load context, check conflicts, show priorities |
+| `/session-end` | Session | 10-step end-of-session housekeeping |
+| `/health-check` | Quality | Project integrity audit and health score |
+| `/rebuild-snapshots` | Session | Regenerate all context snapshots from raw files |
+| `/pipeline-update` | Pipeline | Track entity status transitions |
+| `/outreach-sequence` | Pipeline | Design multi-touch outreach cadences |
+| `/process-call` | Discovery | Post-discovery-call structured processing |
+| `/enrich-entity` | Research | Deep research and dashboard enrichment for an entity |
+| `/synthesise` | Research | Cross-file research synthesis into structured memos |
+| `/critical-reasoning` | Analysis | Pressure-test ideas with 4 lenses: truth, consequences, risks, feasibility |
+| `/decision` | Analysis | Record strategic decisions with full rationale |
+| `/compare-options` | Analysis | Score and compare 2–5 strategic options |
+| `/weekly-report` | Reporting | Generate stakeholder-ready weekly summaries |
+
+### Live Dashboard
+
+- **7 pages**: Overview, Pipeline, Competitors, Decisions, Scoring, Timeline, Research Hub
+- Warm cream/teal editorial design with dark mode
+- Auto-rebuilds from markdown/CSV source files
+- Deploys to Vercel on push
+
+> **Note:** The dashboard is intentionally a simple starting point — a 101-level view of your project. Every project is different, so we'd encourage you to add custom pages, new data flows, and visualisations that fit your specific context. The [Skill Authoring Guide](docs/skill-authoring-guide.md) and [Dashboard Architecture](dashboard/CLAUDE.md) docs explain how to extend both the skills and the dashboard. Ask Claude to help — "add a new dashboard page that shows X" works great.
+
+### Memory & Context System
+
+- **5 persistent memory files** tracking research, discovery, decisions, and scoring
+- **4 pre-computed context snapshots** for fast session loading
+- **Three loading modes**: Fast, Standard, Deep — so you only load what you need
+
+---
+
+## Architecture
+
+```
+my-project/
+├── CLAUDE.md                 # Agent instructions & context loading rules
+├── STATUS.md                 # Multi-agent coordination board
+├── project.config.json       # Project configuration (generated by /onboard)
+├── memory/                   # Persistent strategic memory
+├── context/                  # Pre-computed snapshots for fast loading
+├── templates/                # Standard formats (call-prep, call-notes, entity-teardown)
+├── research/                 # Raw research files (competitors, market, technical)
+├── discovery/                # Customer/stakeholder discovery [optional]
+├── data/                     # CSV/JSON source of truth
+├── skills/                   # Project-level Claude Code skills
+├── dashboard/                # Live web dashboard (Vercel-deployed)
+├── docs/                     # Executive summary, memos, reports
+└── scripts/                  # Utility scripts
+```
+
+---
+
+## What Can You Ask Claude?
+
+Beyond the built-in skills, you can talk to Claude in plain English. The stack gives Claude the context it needs to answer strategically — your research, pipeline, decisions, and hypothesis are all loaded into memory. Here are some things you might say:
+
+### Research & Analysis
+
+> "Here's my meeting transcript from the call with Acme Corp. Can you analyse this against my existing research to see what it means for our current hypothesis?"
+
+> "Pull together everything we know about the competitive landscape and tell me where the biggest white space is."
+
+> "I just found this industry report — read it and update our competitor research with anything new."
+
+> "What are the strongest and weakest parts of our hypothesis right now? What evidence are we missing?"
+
+### Discovery & Pipeline
+
+> "I have 3 new leads from a conference. Add them to the pipeline and draft a personalised outreach sequence for each."
+
+> "Based on all the discovery calls so far, what patterns are emerging? Are we hearing the same pain points?"
+
+> "Prep me for my call with Nexus Payments tomorrow — what do we know about them and what should I ask?"
+
+### Decision-Making
+
+> "We need to decide between building our own data pipeline vs. using a third-party vendor. Set up a comparison with the pros, cons, and scores."
+
+> "Play devil's advocate on our go-to-market strategy. What are we not seeing?"
+
+> "We're about to commit to a product-led growth motion. Pressure-test this before we lock it in."
+
+### Reporting & Status
+
+> "Write a 2-paragraph update I can send to the board summarising where we are this week."
+
+> "What's changed since last Monday? Give me a diff of all research and decisions."
+
+> "How close are we to hitting any of our kill conditions?"
+
+These are just starting points — you can ask Claude anything about your project and it will draw on the full context of your research, pipeline, and decisions to answer.
+
+---
+
+## Project Types
+
+The `/onboard` wizard configures the stack based on your project type:
+
+| Type | Discovery | Pipeline | Dashboard | Best For |
+|------|:---------:|:--------:|:---------:|----------|
+| Market Entry | Yes | Yes | Yes | New market or product evaluation |
+| Growth Strategy | Yes | Yes | Yes | Existing product, new channels or segments |
+| Competitor Research | No | No | Yes | Competitive intelligence and landscape mapping |
+| Product Launch / GTM | Yes | Yes | Yes | Bringing a product or feature to market |
+| Internal Implementation | No | Yes | Yes | Rolling out a system, process, or initiative |
+| Vendor / Partner Evaluation | Yes | Yes | Yes | Selecting tools, platforms, or partners |
+| Due Diligence | Yes | Yes | Yes | M&A, investment, or acquisition evaluation |
+| Business Case | No | No | Optional | Building a case for investment or change |
+| Transformation / Change | Yes | Yes | Yes | Organisational or process transformation |
+| Custom | Choose | Choose | Choose | Anything else |
+
+---
+
+## Requirements
+
+- [Claude Code](https://claude.ai/code) CLI
+- Node.js 18+ (for dashboard)
+- Git
+
+---
+
+## Documentation
+
+- **[Getting Started Guide](docs/getting-started.md)** — step-by-step setup for non-technical users (install, configure, deploy)
+- [Dashboard Deployment](dashboard/DEPLOY.md) — quick reference for deploying to Vercel
+- [Executive Summary Template](docs/executive-summary.md) — master hypothesis and strategy memo
+- [Evidence Grading Rules](docs/memos/evidence-grading.md) — how claims are tagged and verified
+- [Skill Authoring Guide](docs/skill-authoring-guide.md) — build your own skills for the stack
+- [Dashboard Architecture](dashboard/CLAUDE.md) — how the dashboard reads, builds, and renders data
+
+---
+
+## Contributing
+
+Contributions are welcome. Please see [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines on submitting issues, proposing features, and opening pull requests.
+
+---
+
+## Licence
+
+MIT — see [LICENSE](LICENSE).

package/template/STATUS.md

@@ -0,0 +1,34 @@
+# Project Status Board
+
+Lightweight coordination file. Agents update this when starting or completing work to avoid duplication.
+
+**Last updated**: {{DATE}}
+
+---
+
+## In Progress
+
+| Workstream | Item | Agent/Owner | Started | Notes |
+|-----------|------|-------------|---------|-------|
+
+## Completed Recently
+
+| Workstream | Item | Completed | Output |
+|-----------|------|-----------|--------|
+
+Older entries in `docs/output/work-log.md`.
+
+## Blocked / Waiting
+
+| Item | Blocked on | Owner |
+|------|-----------|-------|
+
+---
+
+## Agent Coordination Rules
+
+1. **Before starting work**: Check this file. If someone else is working on the same item, coordinate or pick something else.
+2. **When starting work**: Add a row to "In Progress" with your workstream, item, and start date.
+3. **When finishing work**: Move the row to "Completed Recently" and note the output file(s).
+4. **If blocked**: Add to "Blocked / Waiting" so other agents can help unblock.
+5. **Workstream labels**: `Research`, `Discovery`, `Product`, `Docs`, `Data`, `Setup`

package/template/context/competitor-snapshot.md

@@ -0,0 +1,27 @@
+# Competitor Snapshot — {{PROJECT_NAME}}
+
+**Last updated**: {{DATE}}
+**Regenerate after**: new competitor research, capability map changes, synthesis updates
+
+> This snapshot is orientation for standard tasks. For strategic decisions about competitive positioning, use **Deep Mode**: read all raw files in `research/competitors/` before reasoning.
+
+---
+
+## Competitive Landscape Summary
+
+| Competitor | Category | Overlap | Threat Level | Key Differentiator |
+|-----------|----------|---------|-------------|-------------------|
+
+---
+
+## Key Competitive Gaps
+
+<!-- What do competitors NOT do that represents your opportunity? -->
+
+---
+
+## Competitor Patterns
+
+<!-- Emerging themes across competitors: pricing trends, feature convergence, blind spots -->
+
+Source: `research/competitors/`, `memory/research.md`

package/template/context/market-snapshot.md

@@ -0,0 +1,32 @@
+# Market Snapshot — {{PROJECT_NAME}}
+
+**Last updated**: {{DATE}}
+**Regenerate after**: new market research, TAM updates, trend analysis
+
+---
+
+## Market Sizing
+
+| Metric | Value | Evidence Grade |
+|--------|-------|---------------|
+| TAM | | |
+| SAM | | |
+| SOM (Y1) | | |
+
+---
+
+## Key Market Dynamics
+
+**Tailwinds**:
+-
+
+**Headwinds**:
+-
+
+---
+
+## Market Trends
+
+<!-- Key trends affecting timing and positioning -->
+
+Source: `research/market/`, `docs/memos/reference-context.md`

package/template/context/pipeline-state.md

@@ -0,0 +1,36 @@
+# Pipeline State Snapshot — {{PROJECT_NAME}}
+
+> **OPTIONAL MODULE** — Include this file if your project has an outreach/pipeline component.
+
+**Last updated**: {{DATE}}
+**Regenerate after**: outreach sent, calls booked/completed, {{ENTITY_TYPE_PLURAL}} added/removed
+
+---
+
+## Pipeline Summary
+
+- **Total {{ENTITY_TYPE_PLURAL}}**: 0
+- **Active outreach**: 0
+- **Calls booked**: 0
+- **Calls completed**: 0
+
+---
+
+## Status Breakdown
+
+| Status | Count | {{ENTITY_TYPE_PLURAL}} |
+|--------|-------|------------------------|
+| Not started | | |
+| Contacted | | |
+| Engaged | | |
+| Meeting booked | | |
+| Meeting completed | | |
+| Dead | | |
+
+---
+
+## Next Actions
+
+<!-- What outreach or follow-ups are due? -->
+
+Source: `{{PIPELINE_SOURCE_OF_TRUTH}}`

package/template/context/project-state.md

@@ -0,0 +1,45 @@
+# Project State Snapshot — {{PROJECT_NAME}}
+
+**Last updated**: {{DATE}}
+**Regenerate after**: discovery calls, kill condition changes, strategic decisions, hypothesis reviews
+
+> This snapshot is orientation for standard tasks. For hypothesis review, critical reasoning, or strategic decisions, use **Deep Mode**: read all raw files in `discovery/calls/` and `memory/decisions.md` before reasoning. Snapshots filter; raw files contain disconfirming signals.
+
+---
+
+## Mission & Leading Hypothesis
+
+**Mission**: {{ONE_LINE_DESCRIPTION}}
+**Goal**: {{GOAL}}. Team: {{TEAM}}.
+
+**Leading hypothesis**: {{STRATEGIC_HYPOTHESIS}}
+
+Sources: `memory/decisions.md`, `memory/scoring.md`
+
+---
+
+## Kill Conditions
+
+| KC | Condition | Status | Direction | Key Evidence | Source |
+|----|-----------|--------|-----------|-------------|--------|
+
+---
+
+## Discovery Progress
+
+<!-- Summary of calls completed, key findings, and pattern emerging -->
+
+---
+
+## Current Priority
+
+<!-- What should the next session focus on? -->
+
+---
+
+## Open Risks
+
+| # | Risk | Severity | Mitigation |
+|---|------|---------|-----------|
+
+Source: `docs/executive-summary.md` §7

package/template/dashboard/CLAUDE.md

@@ -0,0 +1,36 @@
+# Dashboard — {{PROJECT_NAME}}
+
+## Architecture
+
+The dashboard is a static site that reads project markdown/CSV files via `build-data.js` and outputs JSON consumed by browser-side JavaScript.
+
+```
+research/competitors/*.md    ──┐
+memory/research.md            │
+memory/decisions.md           │
+memory/scoring.md             ├──→  build-data.js  ──→  dashboard/data/*.json
+docs/executive-summary.md     │     (reads files)       (overview, entities, competitors,
+docs/output/status-blurb.md   │     (outputs JSON)       decisions, scoring, timeline,
+docs/output/work-log.md       │                          research)
+{{PIPELINE_SOURCE_OF_TRUTH}}  ┘
+```
+
+## Source of Truth Rules
+
+1. **CSV is source of truth** for {{ENTITY_TYPE}} data — add {{ENTITY_TYPE_PLURAL}} to `{{PIPELINE_SOURCE_OF_TRUTH}}` first, never only to JSON
+2. **Markdown files are source of truth** for research, status, and kill conditions
+3. **JSON files in `dashboard/data/`** are generated outputs — never edit them directly
+
+## Build & Deploy
+
+- **Build**: `node build-data.js`
+- **Dev**: `npm run dev` (auto-rebuilds on file changes)
+- **Deploy**: Push to `main` — Vercel auto-deploys
+
+## Adding New Pages
+
+1. Create `new-page.html` using the sidebar template from `index.html`
+2. Create `js/new-page.js` with page-specific logic
+3. Add extraction logic to `build-data.js` if new data sources are needed
+4. Add nav link to the sidebar in all HTML files
+5. Rebuild: `node build-data.js`

package/template/dashboard/DEPLOY.md

@@ -0,0 +1,60 @@
+# Dashboard Deployment
+
+Quick reference for deploying the strategy dashboard to Vercel.
+
+## Prerequisites
+
+- Node.js 18+
+- A Vercel account (https://vercel.com) connected to your GitHub
+
+## Local Build
+
+```bash
+cd dashboard
+npm install
+node build-data.js
+```
+
+Open `index.html` in a browser to preview locally.
+
+## Deploy to Vercel
+
+### Option 1: Via Vercel Dashboard
+1. Import your GitHub repository at https://vercel.com/new
+2. Set **Root Directory** to `dashboard`
+3. Build settings are auto-detected from `vercel.json`
+4. Deploy
+
+### Option 2: Via Vercel CLI
+```bash
+npm i -g vercel
+cd dashboard
+vercel
+```
+
+## Auto-Deploy
+
+Once connected, every push to `main` triggers a rebuild. The `/session-end` skill commits and pushes automatically, so your dashboard stays current.
+
+## Configuration
+
+`vercel.json` defines:
+- **buildCommand**: `node build-data.js` — regenerates JSON from markdown/CSV
+- **installCommand**: `npm install` — installs csv-parse and marked
+- **outputDirectory**: `.` — serves from the dashboard root
+
+## Data Pipeline
+
+```
+project root files ──→ build-data.js ──→ dashboard/data/*.json ──→ browser
+```
+
+Build reads (read-only): `memory/*.md`, `docs/**/*.md`, `data/*.csv`, `research/**/*.md`
+Build writes: `dashboard/data/overview.json`, `entities.json`, `competitors.json`, `decisions.json`, `scoring.json`, `timeline.json`, `research.json`
+
+## Privacy
+
+- The deployed dashboard is **publicly accessible** by default
+- Use a private GitHub repo, but note the Vercel URL is still reachable
+- For confidential projects, consider Vercel's password protection (paid feature) or local-only viewing
+- `robots.txt` includes `noindex, nofollow` to discourage search engine indexing