mini-drake v2.0

Research project management plugin. From raw concept or document → structured workspace → investor-ready output.

Available Commands

Claude: when the user asks to research a technology, analyze a market, start a project, compare projects, or generate outputs — use the appropriate command below. Do not wait to be asked explicitly by slash command; trigger the right workflow proactively.

Command	Trigger phrases	What it does
/start-doc	"start project", "intake this doc", "upload file", "analyze document"	Intake a document → extract 13 sections → create project workspace
/start-search	"research concept", "find out about X", "research technology", paste concept text	Take a text concept → ask preferences → fill all sheet fields via web search
/start-search	"deep research", "12-case", "thorough analysis", "no gaps", "investment thesis"	12-case structured methodology with dependency mapping → gap-free basicMemo
/sheet	"generate sheet", "make project sheet", "format for investor"	Read basicMemo + memory → produce structured project-sheet.md
/enrich	"enrich project", "fill gaps", "what's missing", "update project data"	Simulate /sheet → find blank fields → guide user to fill blocking sections
/analyze	"analyze project", "personas", "competitive positioning", "beachhead", "market fit"	18-step methodology: personas → priority axes → competitor map → market fit verdict
/compare	"compare projects", "which project is best", "rank projects", "side by side"	Grep-extract all projects → cache → interactive HTML dashboard (cards / table / scoring)
/export	"export project", "print brief", "PDF brief", "investor brief", "send to committee"	Self-contained HTML brief (print → PDF) in investors / internal / committee format
/prefs	"preferences", "set language", "my settings", "custom rules", "set tone"	Manage per-project preferences: language, detail level, tone, custom search rules

How projects work

Each project lives in {workspace}/{project-slug}/:

{project-slug}/
├── memory/
│   ├── basicMemo.md    ← 13 sections: core knowledge base
│   ├── context.md      ← hot cache: current state, preferences, key facts
│   └── history.md      ← full timeline log with tagged entries
├── outputs/
│   ├── project-sheet.md
│   ├── compare-cache.json
│   └── export-*.html
└── research-log.md

Active project is tracked in {workspace}/active-project.md. All commands default to the active project — no need to re-specify the folder each time.

Command decision guide

User uploads a file          → /start-doc
User pastes concept text     → /start-search (fast) or /start-search (thorough)
Project exists, needs gaps   → /enrich
Project exists, format it    → /sheet
Project exists, analyze it   → /analyze
Multiple projects, compare   → /compare
Project done, share it       → /export
Change settings/language     → /prefs

Typical workflow

/start-search "AI-powered precision agriculture platform"
   ↓
/sheet          → investor-ready project sheet
   ↓
/analyze        → personas + competitive chart
   ↓
/export         → print-ready PDF brief

Or for multiple projects:

/start-doc project-A.pdf
/start-doc project-B.pdf
/compare        → side-by-side dashboard

Skills used

Skill	Used by
project-intake	/start-doc, /start-search
research-v2	/start-search
market-analysis	/analyze
enrich	/enrich
conversation-history	All commands (load context at start, save result at end)