popilot 0.2.0

package/scaffold/.context/oscar/workflows/multi-agent.md

@@ -0,0 +1,82 @@
+# Multi-Agent Coordination (Oscar Module)
+
+> Referenced from orchestrator.md core when delegation/parallel execution is needed
+
+---
+
+## Sequential Deployment
+
+```
+"Danny should look at the data first, then Simon can formulate hypotheses based on the results."
+
+Danny analysis complete → Oscar passes results → Simon formulates hypothesis
+```
+
+## Parallel Deployment (Parallel Task tool calls)
+
+Oscar **invokes multiple Task tools simultaneously in a single response** for actual parallel execution.
+
+### Implementation Method
+
+```markdown
+1. Analyze request → Determine need for parallel execution
+2. Create Tasks per agent (parallel calls in the same message):
+
+   Task #1 (Danny):
+   - subagent_type: "general-purpose"
+   - prompt: "You are 📈 Danny (Analyst).
+     [Refer to .context/agents/analyst.md persona]
+     {specific analysis request}
+     Be careful with heavy tables when running DB queries."
+
+   Task #2 (Rita):
+   - subagent_type: "general-purpose"
+   - prompt: "You are 🎤 Rita (Researcher).
+     [Refer to .context/agents/researcher.md persona]
+     {specific research request}"
+
+3. Collect results and synthesize
+```
+
+### Parallel Execution Triggers
+
+| Keyword/Situation | Parallel Targets | Reason |
+|-------------------|-----------------|--------|
+| `churn`, `attrition`, `why leaving` | Danny + Rita | Quantitative + Qualitative simultaneously |
+| `full analysis`, `comprehensive` | Danny + Rita + Vicky | Multi-angle analysis |
+| `quick assessment`, `ASAP` | All relevant agents | Speed priority |
+| `brainstorm`, `ideas` | All 5 agents | Parallelized party mode |
+
+### Parallel vs Sequential Decision Criteria
+
+| Situation | Execution Mode | Reason |
+|-----------|---------------|--------|
+| 2+ independent analyses | **Parallel** | Time savings |
+| Previous results needed for next task | **Sequential** | Dependencies |
+| Multiple heavy table DB accesses | **Sequential** | Load distribution |
+| Fast response needed (ASAP) | **Parallel** | Speed priority |
+| Deep analysis needed (deep dive) | **Sequential** | Quality priority |
+
+### Result Synthesis Format
+
+```markdown
+🎩 Oscar: Parallel analysis complete. Synthesizing results.
+
+## 📈 Danny Analysis Results
+- {key metrics}
+- {patterns/insights}
+
+## 🎤 Rita Analysis Results
+- {key VOC}
+- {customer psychology}
+
+## 🎩 Oscar Synthesis
+- **Common findings**: {cross-insights}
+- **Next steps**: {follow-up agent suggestions}
+```
+
+### Precautions
+
+1. **Concurrent DB access**: Be careful with simultaneous heavy table queries, specify LIMIT/WHERE in each Task
+2. **MCP rate limit**: Fall back to sequential when accessing the same MCP server concurrently
+3. **Context consistency**: Pass identical baseline information to each Task, specify reference timestamp when synthesizing results

package/scaffold/.context/oscar/workflows/ollie-sage.md

@@ -0,0 +1,128 @@
+# Ollie & Sage Integration (Oscar Module)
+
+> Referenced when `/oscar-loop` or Task decomposition is requested
+
+---
+
+## Ollie (🎩✨) - Task Creator
+
+Oscar's hands-on counterpart. Responsible for breaking down Epics into detailed Stories/Tasks.
+
+### Spawn Triggers
+
+| Trigger | Action |
+|---------|--------|
+| `/oscar-loop` command | Start parallel Ollie spawn |
+| `Epic N, M elaboration` pattern | Multiple Epics → Multiple Ollies |
+| `Break down tasks` pattern | Single Ollie spawn |
+
+### Spawn Method
+
+```markdown
+Task(
+  subagent_type="general-purpose",
+  run_in_background=true,
+  prompt="""
+  You are 🎩✨ Ollie.
+  [Refer to .context/agents/ollie.md persona]
+
+  [Task] Break down Epic {N} into detailed Tasks/Stories
+  [Collaboration] Collaborate with Danny, Rita, Simon as needed
+  [Deliverables] Story list, collaboration log, uncertainties
+  """
+)
+```
+
+### Result Review Criteria
+
+| Confidence | Oscar Action |
+|------------|-------------|
+| ≥ 0.85 | Auto-approve |
+| 0.5 ~ 0.85 | Oscar's discretion |
+| < 0.5 | Sage consultation required |
+
+---
+
+## Sage (🔮) - Strategic Advisor
+
+Oscar's strategic advisor. Provides big-picture direction when judgment is difficult.
+
+### Consultation Triggers
+
+| Situation | Trigger |
+|-----------|---------|
+| Ollie confidence < 0.5 | Automatic consultation request |
+| Oscar confidence < 70% | **Recommended** |
+| 2+ options under consideration | **Recommended** |
+| Sprint goal alignment in question | Required |
+| Unfamiliar domain/situation | **Recommended** |
+| 3+ rejection iterations | Automatic intervention |
+| Potential Guard Rail violation | Automatic warning |
+
+### Consultation Request Method
+
+```markdown
+Task(
+  subagent_type="general-purpose",
+  run_in_background=false,
+  prompt="""
+  You are 🔮 Sage.
+  [Refer to .context/agents/sage.md persona]
+
+  [Situation] {situation description}
+  [Oscar's concern] {what's difficult to decide}
+  [Sprint goal] {current goal}
+
+  [Request]
+  1. Approval/rejection judgment
+  2. Risk identification
+  3. Recommendations
+  """
+)
+```
+
+### Sage Response Handling
+
+- `approve` → Approve result
+- `reject` → Modification needed, apply Sage's feedback
+- `clarify` → Request clarification from user
+
+---
+
+## /oscar-loop Full Flow
+
+```
+User: "/oscar-loop Elaborate on Epic 2, 3"
+              │
+              ▼
+      🎩 Oscar: Analyze request → Identify N independent tasks
+              │
+      ┌───────┴───────┐
+      │               │
+      ▼               ▼
+🎩✨ Ollie #1   🎩✨ Ollie #2
+(Epic 2)         (Epic 3)
+      │               │
+      ▼               ▼
+  Submit result   Submit result
+      │               │
+      └───────┬───────┘
+              │
+      🎩 Oscar: Review results
+              │
+      ┌───────┼───────┐
+   ≥0.85   0.5~0.85  <0.5
+      │       │        │
+   Auto-    Discretion 🔮 Sage consultation
+   approve     │        │
+              └───┬────┘
+                  ▼
+            Final approve/reject → (If rejected, re-run Ollie)
+```
+
+### Precautions
+
+1. **API cost**: Don't spawn more Ollies than necessary
+2. **Context passing**: Specify sprint goals and related document paths for each Ollie
+3. **Circular reference prevention**: Prevent Ollie from spawning another Ollie
+4. **Timeout**: Recommended maximum 300 seconds per Ollie, Oscar intervenes if exceeded

package/scaffold/.context/oscar/workflows/session-git.md

@@ -0,0 +1,71 @@
+# Session-based Git Operations (Oscar Module)
+
+> Referenced when "commit", "push" requests are made
+
+---
+
+## Core Principles
+
+To avoid breaking other sessions' work in a parallel session environment:
+
+1. **Commit only current session files**: Only process files recorded in the "deliverables" section of the session file
+2. **Protect other session files**: Ignore other files even if they appear in `git status`
+3. **Explicit confirmation**: Show the file list to the user and confirm before committing
+
+## Git Request Detection Keywords
+
+| Keyword | Action |
+|---------|--------|
+| `commit`, `save and push` | Selective per-session commit |
+| `push`, `upload` | Commit then push |
+| `full commit`, `all changes`, `--all` | Full commit (only on explicit request) |
+
+## Execution Flow
+
+```
+1. Check current session files
+   → Extract file list from "deliverables" section in sessions/active/{current session}.md
+
+2. Cross-check with git status
+   → Session deliverables ∩ git changed files = commit targets
+
+3. User confirmation
+   → Display "Commit these files?" list
+   → Warn if other session files exist
+
+4. Execute selective commit
+   → git add {session files only}
+   → git commit -m "..."
+   → git push (on request)
+```
+
+## Commit Message Format
+
+```
+feat(oscar): Improve session journal system and authority delegation
+
+- Add automatic session journal generation to /save command
+- Add Oscar proactive delegation principles
+
+Session: {current session ID}
+Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
+```
+
+## Deliverable Tracking
+
+Oscar **automatically updates the deliverables section of the session file** when modifying files:
+
+```markdown
+## Deliverables
+| File | Change | Description |
+|------|--------|-------------|
+| `.context/agents/orchestrator.md` | Modified | Core lightweight refactor |
+```
+
+This list serves as the basis for which files to commit.
+
+## Precautions
+
+1. **Prevent missing deliverables section**: Always update deliverables section when modifying files
+2. **Same file modified by multiple sessions**: Warn when multiple sessions modify the same file, first-to-commit session takes priority
+3. **Full commit**: Requires explicit "full commit", "all files", or "--all", warns about including other session files