wtf-p 0.4.0 → 0.5.0

package/vendors/claude/commands/wtfp/quick.md

@@ -0,0 +1,107 @@
+---
+name: wtfp:quick
+description: Run a small writing fix without full planning overhead
+argument-hint: "[task description]"
+allowed-tools:
+  - Read
+  - Bash
+  - Write
+  - Edit
+  - Glob
+  - Grep
+  - Task
+  - AskUserQuestion
+---
+
+<objective>
+Execute a quick writing task with atomic commits and state tracking but skip optional agents (plan-checker, argument-verifier).
+
+**Orchestrator role:** Parse task, load minimal context, execute directly or spawn writer agent, commit result.
+
+Use for small targeted tasks: fix a paragraph, add a citation, tweak wording, expand a point.
+</objective>
+
+<context>
+Task: $ARGUMENTS (natural language description of what to do)
+
+**Load project state:**
+@.planning/STATE.md
+@.planning/config.json
+</context>
+
+<process>
+
+## 1. Validate Environment
+
+```bash
+ls .planning/ 2>/dev/null || echo "No project — will work on files directly"
+ls paper/*.md 2>/dev/null
+```
+
+## 2. Parse Task Intent
+
+Classify $ARGUMENTS into task type:
+- **fix** — Correct an error, fix citation, resolve contradiction
+- **add** — Add a paragraph, citation, transition, figure reference
+- **revise** — Rewrite a paragraph, strengthen argument, tighten prose
+- **cite** — Add or fix specific citations
+
+## 3. Load Minimal Context
+
+```bash
+MODEL_PROFILE=$(cat .planning/config.json 2>/dev/null | grep -o '"model_profile"[[:space:]]*:[[:space:]]*"[^"]*"' | grep -o '"[^"]*"$' | tr -d '"' || echo "balanced")
+STATE_CONTENT=$(cat .planning/STATE.md 2>/dev/null)
+```
+
+Only load files directly relevant to the task. No full context dump.
+
+## 4. Execute Task
+
+**For simple tasks (fix, cite):** Execute directly in orchestrator — read target file, make edit, verify.
+
+**For complex tasks (add, revise):** Spawn section-writer agent with focused prompt.
+
+| Agent | quality | balanced | budget |
+|-------|---------|----------|--------|
+| section-writer | opus | sonnet | sonnet |
+
+```
+Task(
+  prompt="First, read ~/.claude/agents/wtfp/section-writer.md for your role.\n\n" + focused_prompt,
+  subagent_type="general-purpose",
+  model="{writer_model}",
+  description="Quick: [task summary]"
+)
+```
+
+**Skip:** plan-checker, argument-verifier (quick mode = no verification loop).
+
+## 5. Commit and Update State
+
+```bash
+git add paper/
+git commit -m "[type](quick): [task description]"
+```
+
+Update STATE.md word count if changed.
+
+</process>
+
+<offer_next>
+
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ WTF-P ► QUICK TASK DONE ✓
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+[task summary] — [words changed]
+
+───────────────────────────────────────────
+
+</offer_next>
+
+<success_criteria>
+- [ ] Task executed with minimal context loading
+- [ ] Changes committed atomically
+- [ ] STATE.md word count updated
+- [ ] No unnecessary agent spawning
+</success_criteria>

package/vendors/claude/commands/wtfp/remove-section.md

@@ -1,6 +1,6 @@
 ---
 name: wtfp:remove-section
-description: Remove a future section and renumber subsequent sections
+description: Remove an unwritten section and renumber the rest
 argument-hint: "[section]"
 allowed-tools:
   - Read

package/vendors/claude/commands/wtfp/report-bug.md

@@ -1,6 +1,6 @@
 ---
 name: wtfp:report-bug
-description: Report a bug in WTF-P via GitHub issue
+description: Report a bug via GitHub issue
 allowed-tools:
   - Bash
   - Read

package/vendors/claude/commands/wtfp/request-feature.md

@@ -1,6 +1,6 @@
 ---
 name: wtfp:request-feature
-description: Request a new feature for WTF-P via GitHub issue
+description: Request a new feature via GitHub issue
 allowed-tools:
   - Bash
   - Read

package/vendors/claude/commands/wtfp/research-gap.md

@@ -1,234 +1,222 @@
 ---
 name: wtfp:research-gap
-description: Comprehensive literature/domain research for a section
-argument-hint: "[section]"
+description: Research literature and domain knowledge for a section
+argument-hint: "[section] [--depth=quick|standard|deep]"
 allowed-tools:
   - Read
   - Bash
   - Write
-  - WebFetch
-  - WebSearch
-  - mcp__context7__*
+  - Task
   - AskUserQuestion
 ---
 
+<execution_context>
+@~/.claude/write-the-f-paper/references/research-pitfalls.md
+</execution_context>
+
 <objective>
 Perform comprehensive literature/domain research for a specific section.
 
-Discovers:
-- Key citations needed
-- Standard approaches in the literature
-- Gaps in current research
-- How experts write about this topic
-
-Creates RESEARCH.md with "how experts write this" knowledge.
+**Orchestrator role:** Validate project, scope research with user, resolve model profile, spawn research-synthesizer agent, present results.
 
-Use for literature reviews, methodology justification, positioning your contribution.
+Creates RESEARCH.md with "how experts write this" knowledge, key citations, gaps, and positioning.
 </objective>
 
-<execution_context>
-@~/.claude/write-the-f-paper/workflows/research-gap.md
-@~/.claude/write-the-f-paper/references/research-pitfalls.md
-</execution_context>
+<context>
+Section: $ARGUMENTS (section number or name)
+</context>
 
 <process>
 
-<step name="verify">
-**Verify project exists:**
+## 1. Validate Environment and Resolve Model Profile
 
 ```bash
-[ ! -f .planning/PROJECT.md ] && echo "ERROR: No project found. Run /wtfp:new-paper first" && exit 1
-[ ! -f .planning/ROADMAP.md ] && echo "ERROR: No outline found. Run /wtfp:create-outline first" && exit 1
+[ ! -f .planning/PROJECT.md ] && echo "ERROR: No project. Run /wtfp:new-paper" && exit 1
+[ ! -f .planning/ROADMAP.md ] && echo "ERROR: No roadmap. Run /wtfp:create-outline" && exit 1
 ```
 
-</step>
+```bash
+MODEL_PROFILE=$(cat .planning/config.json 2>/dev/null | grep -o '"model_profile"[[:space:]]*:[[:space:]]*"[^"]*"' | grep -o '"[^"]*"$' | tr -d '"' || echo "balanced")
+```
 
-<step name="load">
-**Load context:**
+| Agent | quality | balanced | budget |
+|-------|---------|----------|--------|
+| research-synthesizer | opus | sonnet | haiku |
 
-- Read PROJECT.md for core argument and field
-- Read ROADMAP.md for section goal
-- Read argument-map.md for claims to support
-- Read sources/literature.md for existing citations
-</step>
+## 1b. Parse Depth Flag
 
-<step name="scope">
-**Define research scope:**
+Parse --depth from $ARGUMENTS:
+
+```bash
+# Extract depth from arguments (--depth=quick, --depth=standard, or --depth=deep)
+DEPTH_ARG=$(echo "$ARGUMENTS" | grep -oE '\-\-depth=(quick|standard|deep)' | cut -d= -f2)
+
+# If no --depth flag, read default from config
+if [ -z "$DEPTH_ARG" ]; then
+  DEPTH=$(cat .planning/config.json 2>/dev/null | grep -o '"depth"[[:space:]]*:[[:space:]]*"[^"]*"' | grep -o '"[^"]*"$' | tr -d '"' || echo "standard")
+else
+  DEPTH="$DEPTH_ARG"
+fi
+```
+
+**Depth mapping:**
+
+| Depth | Behavior |
+|-------|----------|
+| quick | Quick verbal summary - no RESEARCH.md created, no git commit |
+| standard | Standard research (current behavior) - 10-20 sources, RESEARCH.md created |
+| deep | Exhaustive literature review - 20-50 papers, all categories mandatory |
+
+## 2. Define Research Scope
 
 Use AskUserQuestion:
 - header: "Research Focus"
-- question: "What specifically do you need to understand for [section]?"
+- question: "What specifically do you need to understand for this section?"
 - options:
   - "Key citations" — Who are the must-cite authors/papers?
   - "Methodology" — How do others approach this method?
   - "State of field" — What's the current consensus/debate?
   - "Positioning" — How to differentiate from existing work?
-  - "All of the above" — Comprehensive research
 
-</step>
+## 3. Read Context Files
 
-<step name="research">
-**Conduct research:**
+```bash
+PROJECT_CONTENT=$(cat .planning/PROJECT.md)
+ROADMAP_CONTENT=$(cat .planning/ROADMAP.md)
+ARGMAP_CONTENT=$(cat .planning/structure/argument-map.md 2>/dev/null)
+LITERATURE_CONTENT=$(cat .planning/sources/literature.md 2>/dev/null)
+BIB_INDEX=$(node ~/.claude/bin/bib-index.js index references.bib 2>/dev/null || echo "No references.bib")
+
+SECTION_DIR=$(ls -d .planning/sections/${SECTION}-* 2>/dev/null | head -1)
+CONTEXT_CONTENT=$(cat "${SECTION_DIR}"/*-CONTEXT.md 2>/dev/null)
+```
 
-Based on scope, investigate:
+## 3b. Quick Depth: Verbal Summary Only
 
-**Key Citations:**
-- **Foundational:** Use `node ~/.claude/bin/citation-fetcher.js "[topic]" --intent=seminal` to find high-impact papers.
-- **Recent:** Use `node ~/.claude/bin/citation-fetcher.js "[topic]" --intent=recent --year=2023-2026` to find state-of-the-art.
-- **Validation:** Use `node ~/.claude/bin/citation-fetcher.js "query"` to verify specific papers.
+**If DEPTH is "quick":**
 
-**Standard Approaches:**
-- How do others structure similar sections?
-- What's the typical argument flow?
-- What evidence is typically presented?
+Skip agent spawn entirely. Orchestrator provides direct verbal synthesis.
 
-**Gaps:**
-- What's missing in current literature?
-- Where does your work fit?
-- What questions remain unanswered?
+Read available context (PROJECT.md, argument-map, existing literature, CONTEXT_CONTENT) and synthesize a verbal summary directly to user:
 
-**Expert Writing Patterns:**
-- How do top papers in this area write about this?
-- What terminology is standard?
-- What level of detail is expected?
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ WTF-P ► QUICK RESEARCH SYNTHESIS
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
-</step>
+Based on available context for Section {X}:
 
-<step name="create_research">
-**Create RESEARCH.md:**
+[Synthesize key points from PROJECT, argument-map, and any existing literature.md]
 
-```markdown
----
-section: XX-name
-scope: [research scope]
-confidence: [high/medium/low]
----
+Key areas:
+- [Area 1]: [Brief synthesis]
+- [Area 2]: [Brief synthesis]
 
-# Research: [Section Name]
+Suggested citations to consider:
+- [If any mentioned in existing sources]
 
-## Summary
-[2-3 sentence overview of findings]
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
-## Key Citations
+**Note:** Quick research -- no RESEARCH.md written. Run with `--depth=standard` for persistent research.
+```
 
-### Foundational
-| Citation | Contribution | Relevance |
-|----------|--------------|-----------|
-| [Author (Year)] | [What they established] | [Why cite] |
+No RESEARCH.md created. No git commit. Present summary and offer next steps, then exit workflow.
 
-### Recent/High-Impact
-| Citation | Contribution | Relevance |
-|----------|--------------|-----------|
+**If DEPTH is "standard" or "deep":** Continue to Step 4.
 
-### Methodology
-| Citation | Approach | Relevance |
-|----------|----------|-----------|
+## 4. Spawn wtfp-research-synthesizer Agent (standard/deep)
 
-## Standard Approaches
-[How others write about this topic]
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ WTF-P ► RESEARCHING SECTION {X} (depth: {DEPTH})
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+```
 
-- Typical structure: [...]
-- Common arguments: [...]
-- Expected evidence: [...]
+```
+Task(
+  prompt="First, read ~/.claude/agents/wtfp/research-synthesizer.md for your role and instructions.\n\n" + filled_research_prompt,
+  subagent_type="general-purpose",
+  model="{researcher_model}",
+  description="Research Section {X}"
+)
+```
 
-## Field Consensus
-[What's generally agreed upon]
+Research prompt includes: `<research_scope>` with user-selected scope, `<project_context>` with PROJECT + ROADMAP + argument-map, `<existing_literature>` with bib index and literature.md, `<user_decisions>` with CONTEXT_CONTENT, `<output>` with target RESEARCH.md path, `<depth>` with current depth setting.
 
-## Active Debates
-[Where experts disagree]
+**Depth-specific instructions in prompt:**
 
-## Gaps in Literature
-- [Gap 1]: [your opportunity]
-- [Gap 2]: [your opportunity]
+**If DEPTH is "standard":**
+Add `<depth_instruction>Target 10-20 relevant sources across categories. Focus on most impactful papers for each category.</depth_instruction>`
 
-## Your Positioning
-[How your work relates to existing literature]
+**If DEPTH is "deep":**
+Add `<depth_instruction>Perform exhaustive literature review. Target 20-50 papers across all categories. Include systematic field mapping: major research groups, funding bodies, key venues, citation networks. Map the intellectual lineage of the field. Identify ALL competing approaches, not just the main ones. Document methodology variations comprehensively. All 5 categories (foundational, recent, competing, gap, methods) are REQUIRED, not optional.</depth_instruction>`
 
-## Writing Recommendations
-- Terminology to use: [...]
-- Level of detail expected: [...]
-- Citations to definitely include: [...]
-- Arguments to make: [...]
+**Source priority hierarchy (all depths):**
+Add `<source_priority>Source priority: Context7 MCP first (if available), official sources/databases second (Google Scholar, PubMed, IEEE, ACL, arXiv), WebSearch last.</source_priority>`
 
-## Sources Consulted
-- [Source 1]
-- [Source 2]
+## 5. Handle Researcher Return
 
-## Confidence Assessment
-- Key citations: [high/medium/low]
-- Field understanding: [high/medium/low]
-- Gap identification: [high/medium/low]
+**`## RESEARCH COMPLETE`:** Present findings summary.
 
-## Open Questions
-- [Questions that need user input]
+**`## RESEARCH BLOCKED`:** Show blocker, offer alternatives.
 
----
-*Researched: [date]*
-```
+## 6. Present Results
 
-</step>
+</process>
 
-<step name="update_literature">
-**Update literature index:**
+<offer_next>
 
-Add any new citations discovered to `.planning/sources/literature.md`.
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ WTF-P ► RESEARCH COMPLETE ✓
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
-</step>
+Depth: {DEPTH}
+Citations found: {N}
+Confidence: {level}
 
-<step name="commit">
-```bash
-git add .planning/sections/XX-name/XX-RESEARCH.md .planning/sources/literature.md
-git commit -m "$(cat <<'EOF'
-research(XX): [section name] literature review
-
-Scope: [scope]
-Citations found: [N]
-Confidence: [level]
-EOF
-)"
-```
+───────────────────────────────────────────
 
-</step>
+## ▶ Next Up
 
-<step name="done">
-**Present completion:**
+**Plan section** — create writing plan with research context
 
-```
-Research complete:
+`/wtfp:plan-section {section}`
 
-- Research: .planning/sections/XX-name/XX-RESEARCH.md
-- Citations added: [N] new references
-- Confidence: [level]
+<sub>`/clear` first → fresh context window</sub>
 
-## Key Findings
-[2-3 bullet summary]
+───────────────────────────────────────────
 
-## Recommended Next Steps
-[Based on research]
+</offer_next>
 
----
+<offer_next_quick>
+
+**For quick depth only (no RESEARCH.md written):**
+
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+ WTF-P ► QUICK RESEARCH COMPLETE
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+Quick research -- no RESEARCH.md written. Run with `--depth=standard` for persistent research.
+
+───────────────────────────────────────────
 
 ## ▶ Next Up
 
-**Plan section** — create writing plan with research context
+**Plan section** — proceed without literature doc
 
-`/wtfp:plan-section [section]`
+`/wtfp:plan-section {section}`
 
-<sub>`/clear` first → fresh context window</sub>
+**Or deepen research:**
 
----
-```
+`/wtfp:research-gap {section} --depth=standard`
 
-</step>
+───────────────────────────────────────────
 
-</process>
+</offer_next_quick>
 
 <success_criteria>
 - [ ] Research scope defined with user
-- [ ] Key citations identified
-- [ ] Standard approaches documented
-- [ ] Gaps identified
-- [ ] Writing recommendations provided
-- [ ] literature.md updated
-- [ ] All committed to git
+- [ ] Research-synthesizer spawned with full context
+- [ ] RESEARCH.md created with key citations and gaps
+- [ ] User knows next steps
 </success_criteria>

package/vendors/claude/commands/wtfp/resume-writing.md

@@ -1,6 +1,6 @@
 ---
 name: wtfp:resume-writing
-description: Resume work from previous session with full context restoration
+description: Resume writing from a previous session
 allowed-tools:
   - Read
   - Bash