docs: Add handoff notes for JIT template rendering

Detailed notes on the current state of JIT template rendering
implementation, including what's working, current issues, and
next steps for completion.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

Author: jeremymanning

handoff_notes_jit.md

@@ -0,0 +1,54 @@
+# JIT Template Rendering - Handoff Notes
+
+## Current State
+The JIT template rendering foundation is implemented and committed. The main components are:
+
+1. **TemplateMetadata** class in `src/orchestrator/core/template_metadata.py` - tracks template dependencies
+2. **Template analysis** in `YAMLCompiler._analyze_template()` - identifies which step IDs each template depends on
+3. **Task enhancement** - Task class now has `template_metadata` field to store analysis results
+4. **Control flow integration** - Fixed to preserve template metadata for conditional tasks
+
+## What's Working
+- Template dependency analysis correctly identifies dependencies
+- Basic pipelines (like `research_basic.yaml`) work correctly
+- The control system renders templates at runtime using the template manager
+
+## Current Issue
+The remaining issue is with conditional tasks that have both:
+1. Dependencies on other tasks
+2. A condition that uses results from those dependencies
+
+Example from `research_advanced_tools.yaml`:
+```yaml
+- id: extract_content
+  dependencies: [search_topic, deep_search]
+  condition: "{{ (search_topic.results | length > 0) or (deep_search.results | length > 0) }}"
+```
+
+The problem: We can't evaluate the condition until dependencies complete, but the current code tries to evaluate it too early.
+
+## Solution Approach
+The solution is to ensure conditional tasks only evaluate their conditions AFTER their declared dependencies are satisfied. This requires modifying the execution flow in `orchestrator.py`.
+
+## Debugging Commands
+Test the implementation:
+```bash
+# Test basic pipeline (should work)
+python scripts/run_pipeline.py examples/research_basic.yaml -i topic="test"
+
+# Test advanced pipeline (currently fails on conditional task)
+python scripts/run_pipeline.py examples/research_advanced_tools.yaml -i topic="test"
+```
+
+## Key Files to Review
+1. `src/orchestrator/orchestrator.py` - Lines 483-503 where conditional task evaluation happens
+2. `src/orchestrator/control_flow/conditional.py` - The ConditionalTask.should_execute() method
+3. `src/orchestrator/core/control_system.py` - The _render_task_templates() method
+
+## Next Steps
+1. Ensure conditional tasks wait for dependencies before evaluating conditions
+2. Test all example pipelines to ensure they work
+3. Add unit tests for template analysis and rendering
+4. Update documentation
+
+The implementation is very close - just need to fix the timing of conditional task evaluation.