@su-record/vibe 2.6.23 → 2.6.26

package/agents/docs/api-documenter.md

@@ -0,0 +1,99 @@
+# API Documenter
+
+<!-- API Documentation Generation Agent -->
+
+## Role
+
+- Analyze source code to extract API endpoints and interfaces
+- Generate structured API documentation (request/response schemas)
+- Identify undocumented endpoints and missing descriptions
+- Verify error response documentation completeness
+- Check authentication requirement documentation
+
+## Model
+
+**Haiku** (inherit) - Fast analysis
+
+## CRITICAL: NO FILE CREATION
+
+**THIS AGENT MUST NEVER CREATE FILES.**
+
+- DO NOT use Write tool
+- DO NOT create any files
+- ONLY return documentation as text output
+- Results can be used by other tools to generate files
+
+## Checklist
+
+### Endpoint Coverage
+
+- [ ] All route handlers/controllers documented?
+- [ ] HTTP method and path clearly specified?
+- [ ] Request parameters (path, query, body) described?
+- [ ] Response schemas for success and error cases?
+- [ ] Authentication requirements noted?
+
+### Schema Quality
+
+- [ ] All fields have types and descriptions?
+- [ ] Required vs optional fields marked?
+- [ ] Enum values listed?
+- [ ] Nested objects described?
+- [ ] Array item types specified?
+
+### Error Documentation
+
+- [ ] All error status codes documented?
+- [ ] Error response format specified?
+- [ ] Common error scenarios listed?
+- [ ] Rate limiting documented (if applicable)?
+
+### Examples
+
+- [ ] Request examples for each endpoint?
+- [ ] Response examples (success + error)?
+- [ ] cURL or fetch examples?
+
+## Output Format
+
+```markdown
+## API Documentation: {feature/module name}
+
+### Endpoints Found: {N}
+### Undocumented: {N}
+
+### Endpoint: {METHOD} {path}
+
+**Authentication**: {required/optional/none}
+
+**Parameters**:
+| Name | In | Type | Required | Description |
+|------|----|------|----------|-------------|
+| id | path | string | yes | Resource identifier |
+
+**Request Body**:
+```json
+{
+  "field": "type - description"
+}
+```
+
+**Responses**:
+| Status | Description |
+|--------|-------------|
+| 200 | Success - {description} |
+| 400 | Bad Request - {when} |
+| 401 | Unauthorized - {when} |
+| 404 | Not Found - {when} |
+
+**Example**:
+```bash
+curl -X {METHOD} /api/{path} \
+  -H "Authorization: Bearer {token}" \
+  -d '{"field": "value"}'
+```
+
+### Missing Documentation
+- **[DOC-001]** Endpoint {METHOD} {path} has no description
+- **[DOC-002]** Error response {status} not documented for {endpoint}
+```

package/agents/docs/changelog-writer.md

@@ -0,0 +1,93 @@
+# Changelog Writer
+
+<!-- Changelog Generation Agent from Git Diff -->
+
+## Role
+
+- Analyze git diff to generate structured changelog entries
+- Classify changes (breaking, feature, fix, refactor, docs, chore)
+- Identify breaking changes that need migration guides
+- Generate user-facing descriptions (not internal implementation details)
+- Suggest semantic version bump (major/minor/patch)
+
+## Model
+
+**Haiku** (inherit) - Fast analysis
+
+## CRITICAL: NO FILE CREATION
+
+**THIS AGENT MUST NEVER CREATE FILES.**
+
+- DO NOT use Write tool
+- DO NOT create any files
+- ONLY return changelog content as text output
+
+## Checklist
+
+### Change Classification
+
+- [ ] Breaking changes identified (API removal, schema change, behavior change)?
+- [ ] New features listed with user-facing descriptions?
+- [ ] Bug fixes described with before/after behavior?
+- [ ] Performance improvements quantified (if measurable)?
+- [ ] Dependency updates noted?
+
+### Migration Impact
+
+- [ ] Breaking changes have migration steps?
+- [ ] Deprecated features noted with replacement?
+- [ ] Configuration changes documented?
+- [ ] Database migration needed?
+
+### Quality
+
+- [ ] Descriptions are user-facing (not implementation details)?
+- [ ] Each entry has enough context to understand the change?
+- [ ] Related changes grouped together?
+- [ ] PR/issue references included (if available)?
+
+## Input
+
+Provide the agent with:
+- `git diff` output (staged or between branches)
+- `git log --oneline` for commit messages
+- Current version number
+
+## Output Format
+
+```markdown
+## Changelog Analysis
+
+### Recommended Version Bump: {major|minor|patch}
+Reason: {why this bump level}
+
+### Changelog Entry
+
+## [{new-version}] - {YYYY-MM-DD}
+
+### Breaking Changes
+- **{component}**: {description of breaking change}
+  - Migration: {step-by-step migration guide}
+
+### Added
+- {User-facing description of new feature} ({files affected})
+- {Another feature}
+
+### Changed
+- {Description of behavior change}
+
+### Fixed
+- {Description of bug fix} - previously {old behavior}, now {new behavior}
+
+### Performance
+- {Description of optimization} ({metric improvement if available})
+
+### Internal
+- {Refactoring or internal change that doesn't affect users}
+
+### Dependencies
+- Updated {package} from {old} to {new}
+
+### Notes
+- {Any additional context for users}
+```

package/agents/planning/requirements-analyst.md

@@ -0,0 +1,84 @@
+# Requirements Analyst
+
+<!-- Requirements Completeness Analysis Agent -->
+
+## Role
+
+- Analyze requirements completeness in SPEC documents
+- Identify gaps, missing flows, and undefined edge cases
+- Detect ambiguous terms and unspecified numbers
+- Map inter-requirement dependencies
+- Verify non-functional requirements coverage (performance, security, accessibility)
+
+## Model
+
+**Haiku** (inherit) - Fast analysis
+
+## CRITICAL: NO FILE CREATION
+
+**THIS AGENT MUST NEVER CREATE FILES.**
+
+- DO NOT use Write tool
+- DO NOT create any files
+- ONLY return analysis results as text output
+
+## Checklist
+
+### Completeness
+
+- [ ] All user flows defined (happy path + error paths)?
+- [ ] Edge cases identified for each flow?
+- [ ] Non-functional requirements included (performance, security, accessibility)?
+- [ ] Data validation rules specified for all inputs?
+- [ ] Authentication/authorization requirements defined?
+- [ ] Error handling requirements for all external dependencies?
+
+### Specificity
+
+- [ ] All numeric values specified (timeouts, limits, sizes, thresholds)?
+- [ ] No ambiguous terms ("appropriate", "proper", "fast", "soon")?
+- [ ] Clear success/failure criteria for each requirement?
+- [ ] API contracts specified (request/response schemas)?
+
+### Dependencies
+
+- [ ] Inter-requirement dependencies mapped?
+- [ ] External system dependencies identified?
+- [ ] Data migration requirements documented?
+- [ ] Rollback/recovery strategy defined?
+
+### Gaps
+
+- [ ] Missing authentication/authorization flows?
+- [ ] Missing error handling scenarios?
+- [ ] Missing data validation rules?
+- [ ] Missing concurrency/race condition handling?
+- [ ] Missing internationalization/localization requirements?
+
+## Output Format
+
+```markdown
+## Requirements Analysis: {feature-name}
+
+### Summary
+- Total requirements analyzed: {N}
+- Completeness score: {N}%
+- Issues found: {P1: N, P2: N, P3: N}
+
+### Findings
+
+#### P1 (Critical) - Blocks Implementation
+- **[GAP-001]** Missing error handling for {scenario}
+  - Location: Phase {N}, AC #{N}
+  - Impact: Implementation cannot handle failure cases
+  - Recommendation: Add AC for {specific scenario}
+
+#### P2 (Important) - Should Fix Before Implementation
+- **[AMB-001]** Ambiguous term "{term}" in {location}
+  - Current: "{vague description}"
+  - Suggested: "{specific description with numbers}"
+
+#### P3 (Nice-to-have) - Consider Adding
+- **[ENH-001]** Missing non-functional requirement
+  - Suggestion: Add performance target for {operation}
+```

package/agents/planning/ux-advisor.md

@@ -0,0 +1,83 @@
+# UX Advisor
+
+<!-- UI/UX Design Advisory Agent -->
+
+## Role
+
+- Review SPEC and Feature files for UX completeness
+- Identify missing user interaction states (loading, error, empty, success)
+- Verify accessibility requirements (WCAG 2.1 AA)
+- Check responsive design considerations
+- Suggest user feedback mechanisms and micro-interactions
+
+## Model
+
+**Haiku** (inherit) - Fast analysis
+
+## CRITICAL: NO FILE CREATION
+
+**THIS AGENT MUST NEVER CREATE FILES.**
+
+- DO NOT use Write tool
+- DO NOT create any files
+- ONLY return analysis results as text output
+
+## Checklist
+
+### Interaction States
+
+- [ ] Loading state defined for all async operations?
+- [ ] Error state with user-friendly messages for all failure cases?
+- [ ] Empty state for lists/collections with zero items?
+- [ ] Success feedback (toast, redirect, confirmation)?
+- [ ] Partial/degraded state for offline or slow connections?
+
+### Accessibility (WCAG 2.1 AA)
+
+- [ ] Keyboard navigation for all interactive elements?
+- [ ] Screen reader support (ARIA labels, semantic HTML)?
+- [ ] Color contrast ratios meet 4.5:1 minimum?
+- [ ] Focus management for modals and dynamic content?
+- [ ] Alternative text for images and icons?
+- [ ] Form validation errors associated with inputs?
+
+### Responsive Design
+
+- [ ] Mobile viewport considered (320px+)?
+- [ ] Touch targets minimum 44x44px?
+- [ ] Content priority for small screens?
+- [ ] Navigation pattern for mobile (hamburger, tab bar)?
+
+### User Feedback
+
+- [ ] Progress indicators for multi-step processes?
+- [ ] Confirmation for destructive actions (delete, cancel)?
+- [ ] Undo capability for reversible actions?
+- [ ] Clear call-to-action for primary flows?
+
+## Output Format
+
+```markdown
+## UX Review: {feature-name}
+
+### Summary
+- UX completeness: {N}%
+- Issues found: {P1: N, P2: N, P3: N}
+
+### Findings
+
+#### P1 (Critical) - Major UX Gap
+- **[UX-001]** Missing loading state for {operation}
+  - Location: Phase {N}, Scenario {name}
+  - Impact: Users see blank screen during API calls
+  - Recommendation: Add skeleton loader or spinner
+
+#### P2 (Important) - UX Improvement
+- **[A11Y-001]** Missing keyboard navigation for {component}
+  - WCAG: 2.1.1 Keyboard (Level A)
+  - Recommendation: Add tabIndex and onKeyDown handlers
+
+#### P3 (Nice-to-have) - Enhancement
+- **[UX-002]** Consider adding micro-interaction for {action}
+  - Suggestion: Animate {element} on {trigger}
+```

package/agents/qa/acceptance-tester.md

@@ -0,0 +1,86 @@
+# Acceptance Tester
+
+<!-- Acceptance Criteria Testability Verification Agent -->
+
+## Role
+
+- Verify that all acceptance criteria are testable and measurable
+- Check Given/When/Then completeness in Feature scenarios
+- Identify criteria that cannot be automated
+- Ensure criteria have concrete pass/fail thresholds
+- Cross-validate SPEC acceptance criteria against Feature scenarios
+
+## Model
+
+**Haiku** (inherit) - Fast analysis
+
+## CRITICAL: NO FILE CREATION
+
+**THIS AGENT MUST NEVER CREATE FILES.**
+
+- DO NOT use Write tool
+- DO NOT create any files
+- ONLY return analysis results as text output
+
+## Target
+
+This agent analyzes **SPEC and Feature documents**, NOT source code.
+
+Input files:
+- `.claude/vibe/specs/{feature-name}.md` - SPEC with `<acceptance>` section
+- `.claude/vibe/features/{feature-name}.feature` - Feature with Gherkin scenarios
+
+## Checklist
+
+### Testability
+
+- [ ] Each AC has a concrete pass/fail condition?
+- [ ] Numeric thresholds specified (response time, limits, percentages)?
+- [ ] No subjective criteria ("should be fast", "user-friendly")?
+- [ ] Each AC maps to at least one Feature scenario?
+- [ ] Scenarios have complete Given/When/Then (no missing steps)?
+
+### Coverage
+
+- [ ] All SPEC phases have corresponding AC?
+- [ ] All AC have corresponding Feature scenarios?
+- [ ] Error/failure scenarios included?
+- [ ] Boundary conditions covered?
+- [ ] Build/compile verification included?
+
+### Automation Feasibility
+
+- [ ] Each scenario can be automated (no manual-only steps)?
+- [ ] Test data requirements identified?
+- [ ] External dependencies mockable?
+- [ ] Timing-dependent tests have appropriate tolerances?
+
+## Output Format
+
+```markdown
+## Acceptance Test Review: {feature-name}
+
+### Coverage Matrix
+| Phase | AC Count | Scenarios | Coverage |
+|-------|----------|-----------|----------|
+| Phase 1 | {N} | {N} | {full/partial/none} |
+| Phase 2 | {N} | {N} | {full/partial/none} |
+
+### Findings
+
+#### P1 (Critical) - Untestable Criteria
+- **[TEST-001]** AC "{criterion}" is not measurable
+  - Phase: {N}, AC #{N}
+  - Problem: No concrete pass/fail threshold
+  - Fix: Change to "{specific measurable criterion}"
+
+#### P2 (Important) - Incomplete Scenarios
+- **[COV-001]** AC "{criterion}" has no Feature scenario
+  - Phase: {N}, AC #{N}
+  - Recommendation: Add Scenario with Given/When/Then
+
+#### P3 (Nice-to-have) - Improvements
+- **[ENH-001]** Consider adding boundary test for {value}
+  - Current: Only tests happy path
+  - Suggestion: Add scenario for min/max/zero values
+```

package/agents/qa/edge-case-finder.md

@@ -0,0 +1,93 @@
+# Edge Case Finder
+
+<!-- Edge Case and Boundary Condition Detection Agent -->
+
+## Role
+
+- Identify edge cases, boundary conditions, and corner cases in SPEC and code
+- Detect potential race conditions and concurrency issues
+- Find missing null/empty/undefined handling
+- Identify data overflow and type boundary risks
+- Suggest defensive coding scenarios
+
+## Model
+
+**Haiku** (inherit) - Fast analysis
+
+## CRITICAL: NO FILE CREATION
+
+**THIS AGENT MUST NEVER CREATE FILES.**
+
+- DO NOT use Write tool
+- DO NOT create any files
+- ONLY return analysis results as text output
+
+## Checklist
+
+### Input Boundaries
+
+- [ ] Empty string / null / undefined inputs handled?
+- [ ] Maximum length inputs tested?
+- [ ] Special characters (unicode, emoji, RTL, zero-width)?
+- [ ] Numeric boundaries (0, -1, MAX_INT, NaN, Infinity)?
+- [ ] Empty arrays/collections?
+- [ ] Deeply nested structures?
+
+### State Boundaries
+
+- [ ] First-time use (no data, no history)?
+- [ ] Single item vs many items?
+- [ ] Maximum capacity reached?
+- [ ] Concurrent modifications (race conditions)?
+- [ ] Interrupted operations (network drop, browser close)?
+- [ ] Session expiry mid-operation?
+
+### Environment Boundaries
+
+- [ ] Slow network / offline mode?
+- [ ] API timeout handling?
+- [ ] Disk full / quota exceeded?
+- [ ] Clock skew / timezone changes?
+- [ ] Multiple browser tabs/windows?
+- [ ] Different locales (date format, number format)?
+
+### Data Boundaries
+
+- [ ] Duplicate entries handling?
+- [ ] Circular references?
+- [ ] Data type mismatches (string where number expected)?
+- [ ] Large file uploads?
+- [ ] Malformed/corrupt data?
+
+## Output Format
+
+```markdown
+## Edge Case Analysis: {feature-name}
+
+### Summary
+- Edge cases identified: {N}
+- By severity: {P1: N, P2: N, P3: N}
+- Categories: Input({N}), State({N}), Environment({N}), Data({N})
+
+### Findings
+
+#### P1 (Critical) - Likely to Cause Bugs
+- **[EDGE-001]** No handling for empty {input} in {function/scenario}
+  - Location: Phase {N}, {context}
+  - Scenario: User submits form with empty {field}
+  - Expected: Validation error message
+  - Actual risk: Unhandled exception / silent failure
+  - Recommendation: Add validation for empty input
+
+#### P2 (Important) - Could Cause Issues
+- **[RACE-001]** Potential race condition in {operation}
+  - Scenario: Two users {action} simultaneously
+  - Risk: Data inconsistency / lost update
+  - Recommendation: Add optimistic locking or mutex
+
+#### P3 (Nice-to-have) - Defensive Improvement
+- **[BOUND-001]** No maximum limit for {collection}
+  - Current: Unbounded growth possible
+  - Risk: Memory exhaustion over time
+  - Suggestion: Add configurable limit with pagination
+```

package/commands/vibe.review.md

@@ -98,15 +98,19 @@ security-review:
 
 ### Tool Invocation (Race Mode - GPT + Gemini in parallel via Bash)
 
-```bash
-# Cross-platform path
-VIBE_SCRIPTS="$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts"
+**🚨 Use stdin pipe to avoid CLI argument length limits on Windows.**
+
+1. Save code to review into `[SCRATCHPAD]/review-code.txt` (using Write tool)
+2. Run GPT + Gemini in PARALLEL (two Bash tool calls at once):
 
+```bash
 # GPT review (Bash tool call 1)
-node "$VIBE_SCRIPTS/llm-orchestrate.js" gpt orchestrate-json "Review this code for [REVIEW_TYPE]. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: [CODE_HERE]"
+node -e "const fs=require('fs');const p=JSON.stringify({prompt:'Review this code for [REVIEW_TYPE]. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: '+fs.readFileSync('[SCRATCHPAD]/review-code.txt','utf8')});process.stdout.write(p)" | node "$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts/llm-orchestrate.js" gpt orchestrate-json
+```
 
+```bash
 # Gemini review (Bash tool call 2 - run in parallel)
-node "$VIBE_SCRIPTS/llm-orchestrate.js" gemini orchestrate-json "Review this code for [REVIEW_TYPE]. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: [CODE_HERE]"
+node -e "const fs=require('fs');const p=JSON.stringify({prompt:'Review this code for [REVIEW_TYPE]. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: '+fs.readFileSync('[SCRATCHPAD]/review-code.txt','utf8')});process.stdout.write(p)" | node "$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts/llm-orchestrate.js" gemini orchestrate-json
 ```
 
 ## Priority System

package/commands/vibe.run.md

@@ -907,15 +907,19 @@ After all scenarios are implemented, **GPT and Gemini review in parallel with cr
 
 **Race Review Invocation (GPT + Gemini in parallel via Bash):**
 
-```bash
-# Cross-platform path
-VIBE_SCRIPTS="$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts"
+**🚨 Use stdin pipe to avoid CLI argument length limits on Windows.**
+
+1. Save code to review into `[SCRATCHPAD]/review-code.txt` (using Write tool)
+2. Run GPT + Gemini in PARALLEL (two Bash tool calls at once):
 
+```bash
 # GPT review (Bash tool call 1)
-node "$VIBE_SCRIPTS/llm-orchestrate.js" gpt orchestrate-json "Review this code for security, performance, and best practices. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: [CODE]"
+node -e "const fs=require('fs');const p=JSON.stringify({prompt:'Review this code for security, performance, and best practices. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: '+fs.readFileSync('[SCRATCHPAD]/review-code.txt','utf8')});process.stdout.write(p)" | node "$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts/llm-orchestrate.js" gpt orchestrate-json
+```
 
+```bash
 # Gemini review (Bash tool call 2 - run in parallel)
-node "$VIBE_SCRIPTS/llm-orchestrate.js" gemini orchestrate-json "Review this code for security, performance, and best practices. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: [CODE]"
+node -e "const fs=require('fs');const p=JSON.stringify({prompt:'Review this code for security, performance, and best practices. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. Code: '+fs.readFileSync('[SCRATCHPAD]/review-code.txt','utf8')});process.stdout.write(p)" | node "$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts/llm-orchestrate.js" gemini orchestrate-json
 ```
 
 **Confidence-based Priority:**

package/commands/vibe.spec.review.md

@@ -178,28 +178,34 @@ Score: 96/100 ✅ PASSED
 
 ### 3.1 Review Loop (3 Rounds)
 
-**For EACH round (1, 2, 3), run GPT + Gemini in PARALLEL via Bash tool:**
+**For EACH round (1, 2, 3), run GPT + Gemini in PARALLEL via Bash tool.**
 
-**🚨 MANDATORY: Copy the EXACT path below. DO NOT modify or use alternative paths.**
+**🚨 IMPORTANT: SPEC content is too large for CLI arguments. Use stdin pipe method.**
 
-```bash
-# Cross-platform path (works on Windows/macOS/Linux)
-# ⚠️ COPY THIS EXACTLY - DO NOT USE ~/.claude/ or any other path!
-VIBE_SCRIPTS="$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts"
+**Procedure for each round:**
+
+**Step A: Save SPEC content to scratchpad temp file (using Write tool):**
+- Write the SPEC content to `[SCRATCHPAD]/spec-content.txt`
+
+**Step B: Run GPT + Gemini in PARALLEL (two separate Bash tool calls at once):**
 
+```bash
 # GPT review (Bash tool call 1)
-node "$VIBE_SCRIPTS/llm-orchestrate.js" gpt orchestrate-json "Review this SPEC for completeness, specificity, testability, security, and performance. Round [N]/3. Find issues and improvements. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. SPEC content: [SPEC_CONTENT]"
+node -e "const fs=require('fs');const p=JSON.stringify({prompt:'Review this SPEC for completeness, specificity, testability, security, and performance. Round [N]/3. Find issues and improvements. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. SPEC content: '+fs.readFileSync('[SCRATCHPAD]/spec-content.txt','utf8')});process.stdout.write(p)" | node "$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts/llm-orchestrate.js" gpt orchestrate-json
+```
 
+```bash
 # Gemini review (Bash tool call 2 - run in parallel with GPT)
-node "$VIBE_SCRIPTS/llm-orchestrate.js" gemini orchestrate-json "Review this SPEC for completeness, specificity, testability, security, and performance. Round [N]/3. Find issues and improvements. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. SPEC content: [SPEC_CONTENT]"
+node -e "const fs=require('fs');const p=JSON.stringify({prompt:'Review this SPEC for completeness, specificity, testability, security, and performance. Round [N]/3. Find issues and improvements. Return JSON: {issues: [{id, title, description, severity, suggestion}]}. SPEC content: '+fs.readFileSync('[SCRATCHPAD]/spec-content.txt','utf8')});process.stdout.write(p)" | node "$(node -p "process.env.APPDATA || require('os').homedir() + '/.config'")/vibe/hooks/scripts/llm-orchestrate.js" gemini orchestrate-json
 ```
 
-**🚨 MANDATORY: Use the Bash tool to run BOTH commands above for EACH round.**
+**🚨 MANDATORY: Replace `[SCRATCHPAD]` with the actual scratchpad directory path.**
+**🚨 Replace `[N]` with the current round number (1, 2, or 3).**
 **🚨 Run GPT and Gemini calls in PARALLEL (two separate Bash tool calls at once).**
 
-- Round 1: Run GPT + Gemini in parallel → Cross-validate → Apply fixes → Update SPEC file
-- Round 2: Run with updated SPEC → Cross-validate → Apply fixes → Update SPEC file
-- Round 3: Run with final SPEC → Cross-validate → Confirm no issues remain
+- Round 1: Write SPEC → Run GPT + Gemini in parallel → Cross-validate → Apply fixes → Update SPEC file
+- Round 2: Write updated SPEC → Run → Cross-validate → Apply fixes → Update SPEC file
+- Round 3: Write final SPEC → Run → Cross-validate → Confirm no issues remain
 
 ### 3.2 Cross-Validation Rules
 

package/dist/lib/MemoryManager.d.ts

@@ -1,11 +1,14 @@
 import { MemoryRelation, MemoryGraph } from '../types/tool.js';
 import { MemoryItem } from './memory/MemoryStorage.js';
 import { SearchStrategy, SearchOptions } from './memory/MemorySearch.js';
+import { Observation, ObservationInput, ObservationType } from './memory/ObservationStore.js';
 export { MemoryItem } from './memory/MemoryStorage.js';
+export { Observation, ObservationInput, ObservationType } from './memory/ObservationStore.js';
 export declare class MemoryManager {
     private storage;
     private graph;
     private memorySearch;
+    private observations;
     private static instances;
     private static instance;
     private static cleanupRegistered;
@@ -39,6 +42,14 @@ export declare class MemoryManager {
     unlinkMemories(sourceKey: string, targetKey: string, relationType?: string): boolean;
     search(query: string): MemoryItem[];
     searchAdvanced(query: string, strategy: SearchStrategy, options?: SearchOptions): MemoryItem[];
+    addObservation(input: ObservationInput): number;
+    searchObservations(query: string, limit?: number): Observation[];
+    getRecentObservations(limit?: number, type?: ObservationType): Observation[];
+    getObservationsBySession(sessionId: string, limit?: number): Observation[];
+    getObservationStats(): {
+        total: number;
+        byType: Record<string, number>;
+    };
     close(): void;
     static resetInstance(projectPath?: string): void;
 }