feat: ModelRouter integration for cost optimization (#11)

Integrates ModelRouter for cost-optimized model selection. Routes tasks to CHEAP (Haiku), CAPABLE (Sonnet), or PREMIUM (Opus) tiers based on task_type parameter.

Changes:
- Add enable_model_routing and task_type parameters to EmpathyLLM
- Route summarize/classify tasks to Haiku (cheap)
- Route code generation/bug fixing to Sonnet (capable) 
- Route coordination/synthesis to Opus (premium)
- Add 10 new tests for routing integration
- Fix httpx import for test compatibility

Saves ~80-96% on API costs for appropriate tasks.

Author: silversurfer562

.claude/CLAUDE copy.md

@@ -386,10 +386,10 @@ result = storage.store_pattern(
 
 ## 📞 Project Contacts
 
-**Security Questions:** security-team@company.com
-**HIPAA Compliance:** hipaa-officer@company.com
-**Code Review:** tech-lead@company.com
-**General:** empathy-framework@company.com
+**Security Questions:** admin@smartaimemory.com
+**HIPAA Compliance:** admin@smartaimemory.com
+**Code Review:** admin@smartaimemory.com
+**General:** admin@smartaimemory.com
 
 ---
 

.claude/CLAUDE.md

@@ -0,0 +1,42 @@
+---
+paths: **/*.py, **/*.js, **/*.ts
+---
+
+# Bug Patterns (Auto-generated by Empathy)
+Last sync: 2025-12-18 23:11
+
+These patterns help identify and fix common bugs based on your team's history.
+
+## Async Timing Bugs
+
+### When you see: `Promise { <pending> } returned instead of resolved value`
+
+### When you see: `UnhandledPromiseRejection: await undefined`
+
+## Null Reference Bugs
+
+### When you see: `TypeError: Cannot read property 'map' of undefined`
+
+### When you see: `TypeError: Cannot read property 'map' of undefined`
+**Root cause:** API returned null instead of empty array
+**Fix:** Added default empty array fallback: data?.items ?? []
+
+### When you see: `TypeError: Cannot read property 'map' of undefined`
+**Root cause:** API returns undefined instead of empty array
+**Fix:** Added optional chaining and default array fallback
+
+### When you see: `TypeError: Cannot read property 'length' of undefined`
+
+### When you see: `TypeError: Cannot read property 'length' of undefined`
+
+## Unknown Bugs
+
+### When you see: `ReferenceError: fetchUser is not defined`
+
+## Type Mismatch Bugs
+
+### When you see: `TypeError: expected str but got bytes`
+
+## Import Error Bugs
+
+### When you see: `ModuleNotFoundError: No module named 'pandas'`

.claude/rules/empathy/bug-patterns.md

@@ -0,0 +1,39 @@
+---
+paths: **/*
+---
+
+# Coding Patterns (Auto-generated by Empathy)
+Last sync: 2025-12-18 23:11
+
+Coding patterns and quality findings from automated inspection.
+
+## Debt
+
+- # Use temporary directory for test logs
+
+## General
+
+- Historical import_error bugs suggest testing gaps
+- Historical unknown bugs suggest testing gaps
+- Historical import_error bugs suggest testing gaps
+- Historical async_timing bugs suggest testing gaps
+- Historical null_reference bugs suggest testing gaps
+
+## Review
+
+- Potential null/undefined reference
+- Potential null/undefined reference (historical: bug_20250915_abc123)
+- Potential null/undefined reference (historical: bug_20251212_97c0f72f)
+- Potential null/undefined reference (historical: bug_null_001)
+- Potential null/undefined reference (historical: bug_20251212_3c5b9951)
+- Potential null/undefined reference (historical: bug_20251212_a9b6ec41)
+- Potential null/undefined reference (historical: bug_20251212_a0871d53)
+- Potential null/undefined reference (historical: bug_20250822_def456)
+
+## Debugging
+
+- Unresolved bug: null_reference
+
+## Format
+
+- File needs reformatting

.claude/rules/empathy/coding-patterns.md

@@ -0,0 +1,12 @@
+---
+paths: **/*.py, **/*.js, **/*.ts
+---
+
+# Security Decisions (Auto-generated by Empathy)
+Last sync: 2025-12-18 23:11
+
+Team security decisions and accepted risks. Reference these before flagging issues.
+
+## Accepted Risks
+
+- **unknown**:

.claude/rules/empathy/security-decisions.md

@@ -0,0 +1,8 @@
+---
+paths: **/*
+---
+
+# Tech Debt Hotspots (Auto-generated by Empathy)
+Last sync: 2025-12-18 23:11
+
+Areas of the codebase with accumulated technical debt.

.claude/rules/empathy/tech-debt-hotspots.md

@@ -0,0 +1 @@
+# Trigger preview

.github/preview-trigger.md

@@ -5,6 +5,93 @@ All notable changes to the Empathy Framework will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [2.2.10] - 2025-12-18
+
+### Added
+
+**Dev Wizards Web Backend**
+- New FastAPI backend for wizards.smartaimemory.com deployment
+- API endpoints for Memory-Enhanced Debugging, Security Analysis, Code Review, and Code Inspection
+- Interactive dashboard UI with demo capabilities
+- Railway deployment configuration (railway.toml, nixpacks.toml)
+
+### Fixed
+- PyPI documentation now reflects current README and features
+
+---
+
+## [2.2.9] - 2025-12-18
+
+### Added
+
+**Code Inspection Pipeline**
+- **`empathy-inspect` CLI** - Unified code inspection command combining lint, security, tests, and tech debt analysis
+  - `empathy-inspect .` - Inspect current directory with default settings
+  - `empathy-inspect . --format sarif` - Output SARIF 2.1.0 for GitHub Actions/GitLab/Azure DevOps
+  - `empathy-inspect . --format html` - Generate visual dashboard report
+  - `empathy-inspect . --staged` - Inspect only git-staged changes
+  - `empathy-inspect . --fix` - Auto-fix safe issues (formatting, imports)
+
+**SARIF 2.1.0 Output Format**
+- Industry-standard static analysis format for CI/CD integration
+- GitHub code scanning annotations on pull requests
+- Compatible with GitLab, Azure DevOps, Bitbucket, and other SARIF-compliant platforms
+- Proper severity mapping: critical/high → error, medium → warning, low/info → note
+
+**HTML Dashboard Reports**
+- Professional visual reports for stakeholders
+- Color-coded health score gauge (green/yellow/red)
+- Six category breakdown cards (Lint, Security, Tests, Tech Debt, Code Review, Debugging)
+- Sortable findings table with severity and priority
+- Prioritized recommendations section
+- Export-ready for sprint reviews and security audits
+
+**Baseline/Suppression System**
+- **Inline suppressions** for surgical control:
+  - `# empathy:disable RULE reason="..."` - Suppress for current line
+  - `# empathy:disable-next-line RULE` - Suppress for next line
+  - `# empathy:disable-file RULE` - Suppress for entire file
+- **JSON baseline file** (`.empathy-baseline.json`) for project-wide policies:
+  - Rule-level suppressions with reasons
+  - File-level suppressions for legacy code
+  - TTL-based expiring suppressions with `expires_at`
+- **CLI commands**:
+  - `--no-baseline` - Show all findings (for audits)
+  - `--baseline-init` - Create empty baseline file
+  - `--baseline-cleanup` - Remove expired suppressions
+
+**Language-Aware Code Review**
+- Integration with CrossLanguagePatternLibrary for intelligent pattern matching
+- Language-specific analysis for Python, JavaScript/TypeScript, Rust, Go, Java
+- Cross-language insights: "This Python None check is like the JavaScript undefined bug you fixed"
+- No false positives from applying wrong-language patterns
+
+### Changed
+
+**Five-Phase Pipeline Architecture**
+1. **Static Analysis** (Parallel) - Lint, security, tech debt, test quality run simultaneously
+2. **Dynamic Analysis** (Conditional) - Code review, debugging only if Phase 1 finds triggers
+3. **Cross-Analysis** (Sequential) - Correlate findings across tools for priority boosting
+4. **Learning** (Optional) - Extract patterns for future inspections
+5. **Reporting** (Always) - Unified health score and recommendations
+
+**VCS Flexibility**
+- Optimized for GitHub but works with GitLab, Bitbucket, Azure DevOps, self-hosted Git
+- Git-native pattern storage in `patterns/` directory
+- SARIF output compatible with any CI/CD platform supporting the standard
+
+### Fixed
+- Marked 5 demo bug patterns from 2025-12-16 with `demo: true` field
+- Type errors in baseline.py stats dictionary and suppression entry typing
+- Type cast for suppressed count in reporting.py
+
+### Documentation
+- Updated [CLI_GUIDE.md](docs/CLI_GUIDE.md) with full `empathy-inspect` documentation
+- Updated [README.md](README.md) with Code Inspection Pipeline section
+- Created blog post draft: `drafts/blog-code-inspection-pipeline.md`
+
+---
+
 ## [2.2.7] - 2025-12-15
 
 ### Fixed

CHANGELOG.md

@@ -19,6 +19,7 @@ empathy-memory serve
 - **Dual-layer architecture** — Redis for millisecond short-term ops, pattern storage for long-term knowledge
 - **AI that learns across sessions** — Patterns discovered today inform decisions tomorrow
 - **Cross-team knowledge sharing** — What one agent learns, all agents can use
+- **Git-native storage** — Optimized for GitHub, works with any VCS (GitLab, Bitbucket, Azure DevOps, self-hosted)
 
 ### Enterprise-Ready
 - **Your data stays local** — Nothing leaves your infrastructure
@@ -147,6 +148,48 @@ The API server runs at `http://localhost:8765` with endpoints for status, stats,
 
 **VS Code Extension:** A visual panel for monitoring memory is available in `vscode-memory-panel/`.
 
+## Code Inspection Pipeline (New in v2.2.9)
+
+Unified code quality with cross-tool intelligence:
+
+```bash
+# Run inspection
+empathy-inspect .
+
+# Multiple output formats
+empathy-inspect . --format json       # For CI/CD
+empathy-inspect . --format sarif      # For GitHub Actions
+empathy-inspect . --format html       # Visual dashboard
+
+# Filter targets
+empathy-inspect . --staged            # Only staged changes
+empathy-inspect . --changed           # Only modified files
+
+# Auto-fix safe issues
+empathy-inspect . --fix
+
+# Suppress false positives
+empathy-inspect . --baseline-init     # Create baseline file
+empathy-inspect . --no-baseline       # Show all findings
+```
+
+**Pipeline phases:**
+1. Static Analysis (parallel) — Lint, security, debt, test quality
+2. Dynamic Analysis (conditional) — Code review, debugging
+3. Cross-Analysis — Correlate findings across tools
+4. Learning — Extract patterns for future use
+5. Reporting — Unified health score
+
+**GitHub Actions SARIF integration:**
+```yaml
+- run: empathy-inspect . --format sarif --output results.sarif
+- uses: github/codeql-action/upload-sarif@v2
+  with:
+    sarif_file: results.sarif
+```
+
+[Full documentation →](docs/CLI_GUIDE.md#code-inspection-pipeline-new-in-v229)
+
 ## License
 
 **Fair Source License 0.9** - Free for students, educators, and teams ≤5 employees. Commercial license ($99/dev/year) for larger organizations. [Details →](LICENSE)

README.md

@@ -20,17 +20,20 @@
 Licensed under Fair Source 0.9
 """
 
-from .base import AgentConfig, BaseAgent, MemDocsConfig, OpusAgent, RedisConfig, SonnetAgent
+from .base import (
+    AgentConfig,
+    BaseAgent,
+    MemDocsConfig,
+    OpusAgent,
+    RedisConfig,
+    SonnetAgent,
+)
 from .editor_agent import EditorAgent
-from .learning import (
-    # Pattern Extraction
+from .learning import (  # Pattern Extraction; Feedback Loop; Quality Gap Detection; SBAR Handoffs
     ExtractedPattern,
-    # Feedback Loop
     FeedbackEntry,
     FeedbackLoop,
-    # Quality Gap Detection
     GapSeverity,
-    # SBAR Handoffs
     HandoffType,
     PatternExtractor,
     QualityGap,