feat: enforce file paths in agentic outputs with JSON schema validation

Agentic tools were returning abstract component references without file
locations, making it impossible for downstream agents to drill into code.
This change enforces structured outputs with required file_path fields
using JSON Schema validation at the LLM provider level.

- Add comprehensive schemas for all 7 agentic tools in agentic_schemas.rs
- Each schema combines freeform analysis field with structured arrays
- FileLocation type enforces name, file_path, and optional line_number
- Schemas passed to LLM via response_format in GenerationConfig
- AutoAgents adapter converts StructuredOutputFormat to ResponseFormat
- MCP handler parses and surfaces structured_output in responses
- Test script updated to extract and display file locations

Benefits: File paths are mandatory, agents can navigate to exact code,
consistent data structure enables better agent-to-agent collaboration.

Author: jruokola

CHANGELOG.md

@@ -36,12 +36,37 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - **Schema**: HNSW index for `embedding_768` column with EFC 200, M 16
 - **Auto-detection**: Automatic column selection based on embedding dimension
 
-#### **File Location Requirements in Agent Outputs**
-- **All EXPLORATORY prompts** now require file locations in responses
+#### **Structured Output Enforcement with JSON Schemas**
+- **JSON schema enforcement** for all 7 agentic tools with **required file paths**
+- **Schema-driven outputs**: LLM cannot return response without file locations
+- **New module**: `codegraph-ai/src/agentic_schemas.rs` with comprehensive schemas:
+  - `CodeSearchOutput`: analysis + components[] + patterns[]
+  - `DependencyAnalysisOutput`: analysis + components[] + dependencies[] + circular_dependencies[]
+  - `CallChainOutput`: analysis + entry_point + call_chain[] + decision_points[]
+  - `ArchitectureAnalysisOutput`: analysis + layers[] + hub_nodes[] + coupling_metrics[]
+  - `APISurfaceOutput`: analysis + endpoints[] + usage_patterns[]
+  - `ContextBuilderOutput`: comprehensive context with all analysis dimensions
+  - `SemanticQuestionOutput`: answer + evidence[] + related_components[]
+- **Required fields**: Every component must include `name`, `file_path`, and optional `line_number`
+- **Provider integration**:
+  - Added `response_format` field to `GenerationConfig`
+  - OpenAI compatible providers send JSON schema to LLM API
+  - AutoAgents adapter converts `StructuredOutputFormat` to CodeGraph `ResponseFormat`
+- **Hybrid output**: Combines freeform `analysis` field with structured arrays
+- **MCP handler**: Parses structured JSON and surfaces in `structured_output` field
+- **Benefits**:
+  - File paths are **mandatory** - no more abstract references
+  - Downstream tools can navigate directly to relevant code
+  - Consistent data structure for programmatic consumption
+  - Better agent-to-agent collaboration with actionable locations
+
+#### **File Location Requirements in Agent Outputs (Deprecated)**
+- **Superseded by**: Structured output enforcement with JSON schemas (above)
+- **Legacy prompt updates**: All EXPLORATORY prompts requested file locations (now enforced)
 - **Format**: `ComponentName in path/to/file.rs:line_number`
 - **Example**: "ConfigLoader in src/config/loader.rs:42" instead of just "ConfigLoader"
 - **6 prompts updated**: code_search, dependency_analysis, call_chain, architecture, context_builder, semantic_question, api_surface
-- **Enables**: Downstream agents can drill into specific files for detailed analysis
+- **Migration**: Prompts now work in conjunction with schema enforcement
 
 ### 🐛 **Fixed - Critical Database Persistence Bugs**
 
@@ -141,6 +166,14 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - **Ollama support**: Ollama provider now respects chunking configuration
 - **Jina unchanged**: Still uses `JINA_MAX_TOKENS` (provider-specific)
 
+### 📦 **Dependencies**
+
+#### **Added**
+- **schemars** (workspace): JSON schema generation for structured LLM outputs
+  - Used in `codegraph-ai` for agentic schema definitions
+  - Enables compile-time schema validation
+  - Auto-generates JSON Schema from Rust types
+
 ### 📚 **Documentation**
 - **GraphFunctions enrichment plan**: Comprehensive plan saved to `.ouroboros/plans/graphfunctions-enrichment-20251118.md`
   - Schema alignment recommendations

Cargo.lock

@@ -17,6 +17,7 @@ thiserror = { workspace = true }
 anyhow = { workspace = true }
 serde = { workspace = true, features = ["derive"] }
 serde_json = { workspace = true }
+schemars = { workspace = true }
 tracing = { workspace = true }
 parking_lot = { workspace = true }
 uuid = { workspace = true }