flonat-research 0.1.0

package/.context/resources/venue-rankings.md

@@ -0,0 +1,147 @@
+# Venue Rankings Reference
+
+> Global reference for targeting publications — journals and conferences.
+> Journals: aim for **CABS AJG 4 or 4***. Conferences: aim for **CORE A* or A**.
+
+## Ranking Systems
+
+| System | Venue type | Source | Scale | Notes |
+|--------|-----------|--------|-------|-------|
+| **CABS AJG** | Journals | Chartered Association of Business Schools | 1–4* | Most comprehensive UK business journal list |
+| **SJR** | Journals | Scimago / Elsevier | Score + Q1–Q4 | Broad coverage across all disciplines |
+| **FT 50** | Journals | Financial Times | Named list | 50 journals; global MBA ranking input |
+| **CORE** | Conferences | Computing Research and Education (Australasia) | A*, A, B, C | Most widely used for CS/IS conferences |
+
+## Data Files
+
+Static CSV data for programmatic lookups:
+
+| File | Contents | Records |
+|------|----------|---------|
+| `.context/resources/venue-rankings/abs_ajg_2024.csv` | CABS AJG 2024 journal rankings | 1,822 journals |
+| `.context/resources/venue-rankings/core_2026.csv` | CORE ICORE 2026 conference rankings (A*–C) | ~800 conferences |
+| `.context/resources/venue-rankings/CABS-AJG-2024.xlsx` | Original CABS spreadsheet (reference only) | — |
+
+**SJR has no static file** — use the Elsevier Serial Title API for live lookups (requires `SCOPUS_API_KEY`).
+
+### SJR Live Lookup
+
+Query the Elsevier Serial Title API to get SJR score and CiteScore quartile for any journal:
+
+```python
+import httpx, json
+
+async def lookup_sjr(title: str, api_key: str) -> dict | None:
+    """Returns {"sjr": float, "quartile": "Q1"|"Q2"|"Q3"|"Q4"} or None."""
+    r = await httpx.AsyncClient(headers={
+        "X-ELS-APIKey": api_key, "Accept": "application/json",
+    }).get("https://api.elsevier.com/content/serial/title",
+           params={"title": title, "view": "CITESCORE"})
+    if r.status_code != 200:
+        return None
+    entries = r.json().get("serial-metadata-response", {}).get("entry", [])
+    # Match exact title (Elsevier does substring search)
+    norm = lambda s: s.lower().strip().replace(".", "").replace(",", "").replace(":", "").replace("&", "and")
+    entry = next((e for e in entries if norm(e.get("dc:title", "")) == norm(title)), None)
+    if not entry:
+        return None
+    sjr_list = entry.get("SJRList", {}).get("SJR", [])
+    return {"sjr": float(sjr_list[0]["$"]), "quartile": "..."} if sjr_list else None
+```
+
+The Scout app (`research/scout/src/scout/services/rankings.py`) has a full implementation with quartile extraction.
+
+---
+
+## CORE Conference Tiers
+
+| Tier | CORE | Meaning |
+|------|------|---------|
+| Tier 1 | A* | Flagship venue — top of field, highly selective (<20% acceptance) |
+| Tier 2 | A | Excellent venue — strong reputation, competitive (<25% acceptance) |
+| Tier 3 | B | Good venue — solid but less competitive |
+| Tier 4 | C | Acceptable venue — regional or niche |
+
+CORE Portal: https://portal.core.edu.au/conf-ranks/
+
+---
+
+## FT 50 List
+
+> Source: Financial Times (https://www.ft.com/content/3405a512-5cbb-11e1-8f1f-00144feabdc0)
+> Journals marked * were added in recent revisions.
+
+1. Academy of Management Journal
+2. Academy of Management Review
+3. Accounting, Organizations and Society
+4. Administrative Science Quarterly
+5. American Economic Review
+6. Contemporary Accounting Research
+7. Econometrica
+8. Entrepreneurship Theory and Practice
+9. Harvard Business Review
+10. Human Relations*
+11. Human Resource Management
+12. Information Systems Research
+13. Journal of Accounting and Economics
+14. Journal of Accounting Research
+15. Journal of Applied Psychology
+16. Journal of Business Ethics
+17. Journal of Business Venturing
+18. Journal of Consumer Psychology
+19. Journal of Consumer Research
+20. Journal of Finance
+21. Journal of Financial and Quantitative Analysis
+22. Journal of Financial Economics
+23. Journal of International Business Studies
+24. Journal of Management*
+25. Journal of Management Information Systems*
+26. Journal of Management Studies
+27. Journal of Marketing
+28. Journal of Marketing Research
+29. Journal of Operations Management
+30. Journal of Political Economy
+31. Journal of the Academy of Marketing Science*
+32. Management Science
+33. Manufacturing and Service Operations Management*
+34. Marketing Science
+35. MIS Quarterly
+36. Operations Research
+37. Organization Science
+38. Organization Studies
+39. Organizational Behavior and Human Decision Processes
+40. Production and Operations Management
+41. Quarterly Journal of Economics
+42. Research Policy*
+43. Review of Accounting Studies
+44. Review of Economic Studies*
+45. Review of Finance*
+46. Review of Financial Studies
+47. Sloan Management Review
+48. Strategic Entrepreneurship Journal*
+49. Strategic Management Journal
+50. The Accounting Review
+
+---
+
+---
+
+## Conference Metadata Template
+
+When targeting a conference in `/init-project-research`, capture:
+
+```markdown
+## Conference Target
+- **Conference:** <full name> (<acronym>)
+- **CORE ranking:** <A*/A/B/C>
+- **Submission deadline:** <date>
+- **Notification date:** <date>
+- **Camera-ready date:** <date>
+- **Conference dates:** <dates>
+- **Location:** <city, country>
+- **Page limit:** <N pages + refs>
+- **Format:** <LaTeX template / style file>
+- **Review type:** <double-blind / single-blind / open>
+- **Anonymisation required:** <yes/no>
+- **CfP link:** <URL>
+```

package/.context/workflows/README.md

@@ -0,0 +1,69 @@
+# Workflows Guide
+
+> How to use the workflow files in this folder.
+
+## What Are Workflows?
+
+Workflows are step-by-step process guides for recurring tasks. They tell Claude how to help you with specific activities.
+
+**Note:** Some capabilities are now in `skills/` instead — skills are more comprehensive and include prompt templates, while workflows are simpler process guides.
+
+## Available Workflows
+
+| Workflow | When to Use | Trigger Phrase |
+|----------|-------------|----------------|
+| `daily-review.md` | Start of workday | "Plan my day" |
+| `weekly-review.md` | End of week | "Weekly review" |
+| `meeting-actions.md` | After meetings | "Extract actions from my meeting with [name]" |
+| `replication-protocol.md` | Replicating a paper's results | "Help me replicate [paper]" |
+
+## Related Skills
+
+These capabilities are in `skills/` folder (more comprehensive):
+
+| Skill | When to Use | Trigger Phrase |
+|-------|-------------|----------------|
+| `project-safety/` | Starting research projects | "Set up a new project safely" |
+| `code-archaeology/` | Revisiting old code | "Audit this codebase" |
+| `literature/` | Literature search & synthesis | "Build a literature review on [topic]" |
+
+## How to Use
+
+### Natural Language (Recommended)
+Just ask naturally:
+
+> "Help me plan my day"
+> "Extract action items from yesterday's meeting with [Supervisor]"
+
+### Direct Reference
+Point to the specific workflow:
+
+> "Read `.context/workflows/daily-review.md` and help me plan"
+
+## Workflow Summaries
+
+### Daily Review
+**Purpose:** Plan your day with questions, not task dumps
+**Process:** Energy check → Surface data → Prioritise → Create plan
+**Output:** Must Do / Should Do / Could Do list
+
+### Weekly Review
+**Purpose:** Reflect on the week and plan the next
+**Process:** Clear decks → Review completed → Plan Big 3 → Check projects
+**Output:** Week summary + next week priorities
+
+### Meeting Actions
+**Purpose:** Extract tasks from meeting transcripts/notes
+**Process:** Find transcript → Identify action items → Create vault tasks
+**Output:** Tasks in vault with proper attribution
+
+### Replication Protocol
+**Purpose:** Replicate results from a published paper before extending
+**Process:** Inventory targets → Line-by-line translation → Programmatic comparison → Extend
+**Output:** `replication-targets.md` + `replication-report.md` in project directory
+
+## Tips
+
+- **Workflows = processes** — Step-by-step guides for recurring tasks
+- **Skills = capabilities** — Comprehensive instructions with templates
+- **Combine as needed** — E.g., use code-archaeology skill with project-safety skill

package/.context/workflows/daily-review.md

@@ -0,0 +1,91 @@
+# Daily Review Workflow
+
+> How AI should help the user plan their day.
+
+## Approach: Questions First, Then Plan
+
+Don't just generate a task list. Instead, follow this process:
+
+### Step 1: Orientation Questions
+
+Ask the user these questions (adapt based on context):
+
+1. **Energy check:** "How are you feeling today - high energy for deep work, or better suited for lighter tasks?"
+
+2. **Constraints:** "Any meetings or hard commitments today that I should plan around?"
+
+3. **Continuity:** "What were you working on yesterday? Want to continue, or switch focus?"
+
+4. **Pressure points:** "Anything weighing on your mind or feeling overdue?"
+
+### Step 2: Surface Relevant Information
+
+After understanding context, present:
+
+1. **Overdue tasks** from vault (`tasks/`)
+2. **Due today** items
+3. **Upcoming deadlines** (next 7 days)
+4. **Recent meeting action items** not yet addressed
+5. **Current focus** from `current-focus.md`
+
+### Step 3: Help Prioritise
+
+Based on the user's answers and the data:
+
+1. Suggest a **realistic** number of tasks (usually 3-5)
+2. Flag any **blocking** items (where others are waiting)
+3. Consider the **project mix** - avoid all tasks from one project
+4. Account for **transition costs** between different types of work
+
+### Step 4: Create the Plan
+
+Format the daily plan as:
+
+```markdown
+## Today's Focus: [Date]
+
+### Must Do (non-negotiable)
+1. [Task] - [Why it's a must]
+
+### Should Do (important but flexible)
+2. [Task]
+3. [Task]
+
+### Could Do (if time permits)
+4. [Task]
+
+### Parking Lot
+- [Tasks explicitly deferred to another day]
+```
+
+## Example Interaction
+
+**AI:** Good morning! Before we plan your day, a few quick questions:
+- How's your energy today?
+- Any meetings or fixed commitments?
+- Where did you leave off yesterday?
+
+**the user:** Tired. Meeting with [Supervisor] at 2pm. Was working on the journal revision.
+
+**AI:** Got it. Given you're tired and have a meeting at 2pm, I'd suggest:
+
+**Must Do:**
+1. Prepare talking points for [Supervisor] meeting (30 min before)
+
+**Should Do:**
+2. Continue journal revision - Section 4 where you left off
+3. Reply to [Supervisor]'s email about the project timeline
+
+**Could Do:**
+4. Review vault tasks for captured items
+
+That leaves your morning for focused writing and gives you buffer before the meeting. Does this feel right?
+
+---
+
+## Integration Points
+
+- **Vault:** Query `tasks/` for overdue/due items via taskflow MCP
+- **Calendar:** Check for meetings (if connected)
+- **Current Focus:** Read `.context/current-focus.md`
+- **Meeting Notes:** Check recent `@[Date]` pages for unprocessed action items

package/.context/workflows/meeting-actions.md

@@ -0,0 +1,108 @@
+# Meeting Action Item Extraction
+
+> How to process the user's meeting transcripts and extract action items.
+
+## Recording System: Minutes
+
+**Tool:** [minutes](https://github.com/silverstein/minutes) — local-first meeting capture with whisper.cpp transcription, speaker diarization, and structured output.
+
+**Output location:** `~/meetings/` (multi-speaker recordings), `~/meetings/memos/` (voice memos)
+
+**Output format:** Markdown with YAML frontmatter containing:
+- `title`, `date`, `duration`, `type` (meeting/memo)
+- `attendees`, `speaker_map` (diarized speakers → real names)
+- `action_items` (structured: assignee, task, due, status)
+- `decisions` (structured: text, topic)
+
+**Audio routing:** BlackHole 2ch virtual audio device for system audio (Zoom, Meet, Teams). Built-in mic for in-person.
+
+## Meeting Lifecycle Skills
+
+| Step | Skill | When |
+|------|-------|------|
+| Prepare | `/minutes prep` | Before a call — builds relationship brief from prior meetings |
+| Record | `/minutes record` | During meeting — `minutes record` / `minutes stop` |
+| Note | `/minutes note` | During meeting — add timestamped annotations |
+| Debrief | `/minutes debrief` | After meeting — compare outcomes to prep, track decisions |
+| Daily recap | `/minutes recap` | End of day — digest all meetings |
+| Weekly | `/minutes weekly` | End of week — themes, decision arcs, stale commitments |
+| Search | `/minutes search` | Anytime — find past discussions by topic, person, decision |
+
+## Extraction Rules
+
+### What Counts as an Action Item
+
+Look for:
+1. **Explicit commitments:** "I'll do X", "I'm going to...", "I need to..."
+2. **Requests:** "Can you...", "Could you send...", "Please..."
+3. **Agreed next steps:** "The next step is...", "We agreed to..."
+4. **Deadlines mentioned:** "by Friday", "before the meeting", "by end of month"
+
+### What to Capture (Full Context)
+
+For each action item, extract:
+
+| Field | Description | Example |
+|-------|-------------|---------|
+| **Task** | What needs to be done | "Send updated literature review" |
+| **Assignee** | Who should do it (the user or someone else) | the user |
+| **Deadline** | When it's due (if mentioned) | "by next Tuesday" |
+| **Related Project** | Which project this relates to | Journal Revision |
+| **Source Meeting** | Path to the transcript file | `~/meetings/2026-03-29-weekly-standup.md` |
+| **Context** | Why this matters / what was discussed | "Reviewer 2 requested more references on cognitive load" |
+
+### Output Format
+
+Create tasks in the vault's `tasks/` directory as markdown files with YAML frontmatter:
+
+```yaml
+---
+title: "[Action verb] [Object] - [Brief context]"
+status: not-started
+priority: [Infer from urgency/deadline]
+due: YYYY-MM-DD
+project: [project-slug]
+tags: [meeting-action]
+---
+
+- **Context:** [Why this task exists]
+- **From meeting:** [Date] with [Person]
+- **Related to:** [Project]
+```
+
+## Processing Workflow
+
+1. **Check for new transcripts** — `minutes list` or browse `~/meetings/`
+2. **Run debrief** — `/minutes debrief` for structured analysis
+3. **Extract action items** — from YAML frontmatter `action_items:` or transcript scan
+4. **Create in vault** — add to `tasks/` directory with proper frontmatter
+5. **Check for conflicts** — `/minutes debrief` flags decision conflicts with prior meetings
+
+## Special Cases
+
+### Supervisor Action Items
+- Flag these as higher priority by default
+- Tag with the relevant university (relevant institution)
+
+### Research-Related Actions
+- Link to the relevant paper project
+- Consider impact on PhD timeline
+
+### Administrative Actions
+- Often have hard deadlines (forms, claims, bookings)
+- Tag with "Claim" type if it's a reimbursement/refund
+
+## Integration
+
+- **Auto-extraction:** Minutes extracts `action_items` into YAML frontmatter (when LLM summarization is configured)
+- **Manual extraction:** Run `/minutes debrief` or use `meeting-analyst` agent for cross-meeting synthesis
+- **Vault sync:** Tasks are written directly to vault `tasks/` files
+- **Triage:** During daily review, assign priorities and dates
+
+## Cross-Meeting Intelligence
+
+The `meeting-analyst` agent handles questions spanning multiple meetings:
+- Person profiles: "What does X usually bring up?"
+- Decision tracking: "What have we decided about pricing?"
+- Stale commitments: "What's still outstanding?"
+- Preparation: "Prepare me for my call with the Acme team"

package/.context/workflows/replication-protocol.md

@@ -0,0 +1,155 @@
+# Replication Protocol
+
+> Step-by-step process for replicating results from a published paper before extending.
+
+## When to Use
+
+- Replicating a paper's empirical results as a foundation for your own work
+- Verifying a coauthor's code produces the claimed results
+- R&R where a referee asks you to demonstrate you can replicate a benchmark
+
+## Principles
+
+1. **Replicate first, extend later** — never "improve" during replication
+2. **Line-by-line translation** — match the original code's logic exactly
+3. **Programmatic comparison** — no eyeballing; use tolerances
+4. **Document everything** — every discrepancy gets recorded
+
+## Phase 1: Inventory Gold Standard Numbers
+
+Before writing any code:
+
+1. Open the original paper
+2. Extract **every** numerical result you plan to replicate:
+   - Point estimates (coefficients, means, treatment effects)
+   - Standard errors / confidence intervals
+   - Sample sizes (N)
+   - Test statistics (t-stats, F-stats, p-values)
+   - Summary statistics from descriptive tables
+3. Record them in `replication-targets.md`:
+
+```markdown
+# Replication Targets
+
+Source: [Author (Year), "Title"]
+
+## Table 1: Summary Statistics
+| Variable | Paper Value | Our Value | Match? |
+|----------|-------------|-----------|--------|
+| Mean income | 45,230 | | |
+| N | 12,500 | | |
+
+## Table 2: Main Results
+| Specification | Estimate | SE | Our Estimate | Our SE | Match? |
+|--------------|----------|-----|-------------|--------|--------|
+| OLS baseline | 0.034 | 0.012 | | | |
+| IV | 0.051 | 0.018 | | | |
+```
+
+## Phase 2: Line-by-Line Translation
+
+1. Obtain the original code (replication package, GitHub, or request from authors)
+2. Translate to your language (R or Python) **without modifying the logic**:
+   - Same variable names where possible
+   - Same sample restrictions
+   - Same model specifications
+   - Same standard error clustering
+3. Comment each block referencing the original code line numbers
+
+**Do NOT:**
+- Fix "bugs" in the original code
+- Use "better" estimators
+- Add robustness checks
+- Clean the code style
+
+## Phase 3: Programmatic Comparison
+
+Compare your results against Phase 1 targets using these tolerances:
+
+| Metric | Tolerance | Rationale |
+|--------|-----------|-----------|
+| Integers (N, counts) | Exact match | No reason for any difference |
+| Point estimates | < 0.01 | Numerical precision differences |
+| Standard errors | < 0.05 | Clustering/bootstrap can vary slightly |
+| p-values | < 0.01 | Derived from above |
+| R-squared | < 0.001 | Numerical precision |
+
+Generate a comparison programmatically:
+
+```r
+# R example
+compare_results <- function(paper_value, our_value, tolerance) {
+  diff <- abs(paper_value - our_value)
+  list(
+    paper = paper_value,
+    ours = our_value,
+    diff = diff,
+    match = diff <= tolerance
+  )
+}
+```
+
+```python
+# Python example
+def compare_results(paper_value, our_value, tolerance):
+    diff = abs(paper_value - our_value)
+    return {
+        "paper": paper_value,
+        "ours": our_value,
+        "diff": diff,
+        "match": diff <= tolerance
+    }
+```
+
+## Phase 4: Extend
+
+Only after Phase 3 produces a clean replication report:
+
+1. Create a **separate script** for extensions (never modify replication code)
+2. Document what you're changing and why
+3. Compare extended results against replication baseline
+
+## Output Files
+
+At the end of a replication, the project directory should contain:
+
+```
+project/
+├── replication-targets.md    ← Phase 1: gold standard numbers
+├── replication-report.md     ← Phase 3: comparison results
+├── code/
+│   ├── 01-replicate.R        ← Phase 2: line-by-line translation
+│   └── 02-extend.R           ← Phase 4: your extensions
+└── data/
+    └── ...
+```
+
+### Replication Report Format
+
+```markdown
+# Replication Report
+
+**Paper:** [Author (Year), "Title"]
+**Date:** YYYY-MM-DD
+**Replicated by:** the user
+
+## Summary
+- Tables replicated: X/Y
+- Figures replicated: X/Y
+- Overall: PASS / PARTIAL / FAIL
+
+## Detailed Results
+[Table-by-table comparison with match status]
+
+## Discrepancies
+[Any values outside tolerance, with investigation notes]
+
+## Notes
+[Software versions, data access issues, ambiguities in original code]
+```
+
+## Cross-References
+
+- **`/code-review`** — Run on replication scripts before finalising
+- **Referee 2 agent** — For formal verification of the full replication
+- **`/code-archaeology`** — If the original replication package needs understanding first

package/.context/workflows/weekly-review.md

@@ -0,0 +1,113 @@
+# Weekly Review Workflow
+
+> Template for the user's weekly reflection and planning.
+
+## When
+
+Ideally: Friday afternoon or Sunday evening
+Duration: 30-45 minutes
+
+## The Review Process
+
+### Part 1: Clear the Decks (10 min)
+
+- [ ] Process inbox (vault tasks with no due date)
+- [ ] Review all "Waiting" status tasks - still waiting?
+- [ ] Check for any loose paper notes, email drafts, or mental open loops
+- [ ] Update `current-focus.md` with where things stand
+
+### Part 2: Review the Week (10 min)
+
+Reflect on:
+
+1. **What got done?**
+   - List completed tasks (check vault tasks with "Done" status)
+   - Note any significant progress on papers/projects
+   - Acknowledge wins, even small ones
+
+2. **What didn't happen?**
+   - Tasks that rolled over multiple times - why?
+   - Commitments made but not kept
+   - Is there a pattern?
+
+3. **What emerged?**
+   - New projects or opportunities
+   - Unexpected challenges
+   - Ideas worth capturing
+
+### Part 3: Plan Next Week (15 min)
+
+1. **Review calendar**
+   - What meetings are scheduled?
+   - Any deadlines?
+   - Travel or unusual commitments?
+
+2. **Identify the BIG 3**
+   - What are the three most important outcomes for next week?
+   - These should move meaningful projects forward
+   - Be realistic given the calendar
+
+3. **Prep for Monday**
+   - What's the first task you'll do?
+   - Any prep needed before a Monday meeting?
+
+### Part 4: Projects Check (10 min)
+
+For each active project (see `projects/_index.md`):
+
+- What's the current status?
+- What's the next action?
+- Any blockers?
+- Timeline still realistic?
+
+## Weekly Review Template
+
+```markdown
+# Week of [DATE]
+
+## Wins
+-
+
+## Challenges
+-
+
+## Key Learnings
+-
+
+## Next Week: Big 3
+1.
+2.
+3.
+
+## Project Status Updates
+| Project | Status | Next Action |
+|---------|--------|-------------|
+| | | |
+
+## Open Questions
+-
+
+## Notes for Future Self
+-
+```
+
+## AI Assistance
+
+When helping with weekly review, AI should:
+
+1. **Pull data from vault**
+   - Completed tasks this week
+   - Overdue tasks
+   - Tasks created this week
+
+2. **Prompt reflection**
+   - "I see you completed X, Y, Z. What felt good about this week?"
+   - "These tasks rolled over - want to reschedule or remove them?"
+
+3. **Help with planning**
+   - "Based on your current projects, here are candidates for Big 3..."
+   - "You have [N] meetings next week - factor this into planning"
+
+4. **Update context files**
+   - Offer to update `current-focus.md` based on the review
+   - Flag any stale information in project files