Visual overview of pipeline stages, team structure, skill routing. Use when onboarding.
Display a comprehensive visual reference of the entire research workflow. Read project state to fill in the current status section.
PIPELINE_STATE.md (if it exists) for the current pipeline stagequality_reports/plans/*.md for the most recent active planPrint the following sections verbatim, replacing only the [dynamic] placeholders in
Section 6 with actual values from the project.
Stage 0 Stage 1 Stage 2 Stage 3 Stage 4 Stage 5 Stage 6
DATA_AUDIT → QUESTIONING → DATA_COLLECTION → ANALYSIS → WRITING → REVIEW → FINAL_APPROVAL
/data-audit /idea-ranking conditions-protocol /stata-... /econ-paper advisor panel (human gate)
🔴 QUESTION_ /research-plan /python-... /tex-journal /paper-tournament
SELECTED gate /structural-estimation /journal-review
🔴 FINAL_APPROVAL
"I need to..." → Use this skill
─────────────────────────────────────────────────────────
Start a new project → /data-audit → /idea-ranking → /research-plan
Find related papers → /lit-review
Write/revise the paper → /econ-paper-writing or /tex-journal-draft
Run estimation code → /stata-reduced-form, /python-econ-data, /structural-estimation
Check my paper before submission → /econ-paper-review (quick) or /journal-review (full)
Get feedback on methodology → /devils-advocate
Work with a coauthor team → /research-team (launches PI + 3 coauthors + specialists)
Respond to referee reports → /revision-plan
Benchmark against published papers → /paper-tournament
Check session health → /context-status
Record a lesson → /learn
Scan an existing project → /research-repo-scanner
Onboard existing project to template → /migrate-project
See this help page → /workflow-help
┌─────────────────┐
│ PI Coordinator │
│ (orchestrates) │
└───────┬─────────┘
┌─────────────┼─────────────┐
▼ ▼ ▼
┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌ ─ ─ ─ ─ ─ ─ ┐
│Econometrician│ │ Paper Writer │ │ Theorist │ Specialist(s)
│ Wave 1 │ │ Wave 2 │ │ Wave 2 │ │ (dynamic) │
│ │ │ │ │ │ Wave 1 or 2
│ scripts/ │ │ paper/ │ │ paper/theory│ │ literature/ │
│ stata/ R/ │ │ slides/ │ │ src/model │ (domain)
│ results/ │ │ literature/ │ │ literature/ │ └ ─ ─ ─ ─ ─ ─ ┘
│ literature/ │ │ │ │ │
│ methods/ │ │ │ │ │
└─────────────┘ └─────────────┘ └─────────────┘
│ │ │
└────────────────┼───────────────┘
▼
Sync Meeting
MERGE / REVISE / REJECT
All coauthors have WebSearch + WebFetch for inline literature searches (3-5 papers).
Broader reviews → PI assigns /lit-review. Specialists are created on-the-fly by the PI.
literature/
├── references.bib ← canonical BibTeX (append-only in team mode)
├── papers/ ← one note per paper ({author}{year}_{keyword}.md)
│ └── _template.md ← copy this for new papers
├── topics/ ← thematic overviews (link to paper notes)
└── methods/ ← estimator/technique notes
Relevance: 1=tangential 2=related 3=important 4=key predecessor 5=direct competitor
Status: unread → reading → read → cited
Team write access:
Econometrician → methods/, papers/ (method-tagged)
Paper Writer → papers/ (empirical/policy), topics/
Theorist → papers/ (theory/mechanism), topics/
Specialist → PI-assigned per domain
Score Meaning Action
───── ────────── ─────────────────
95+ Excellence Aspirational target
90+ PR-ready gh pr create
80+ Commit-ready git commit allowed
<80 Blocked Fix issues first
Deductions: broken code (−100), NaN/Inf (−30), wrong SE clustering (−20),
missing seed (−10), hardcoded paths (−20), unresolved citations (−15)
Read from project files and fill in:
Current stage: [read from PIPELINE_STATE.md → Current Stage, or "not initialized"]
Active plan: [most recent .md in quality_reports/plans/, or "none"]
Human gates: [QUESTION_SELECTED: logged/pending] [FINAL_APPROVAL: logged/pending]