Update TASKS.md and AGENTS.md with comprehensive documentation

Part 2 deliverables from subagent analysis:

## Task Grading (0.00-10.00)

- B.5 ANSI Sanitization: 6.75/10 - Works for common cases but regex
  misses ~30% of ANSI sequences (cursor control, OSC sequences)
- B.4 Content-Block Arrays: 6.75/10 - Handles text/thinking but not
  images or tool_use blocks
- A.1 Copy Buttons: 6.75/10 - Functional but lacks accessibility
  (no ARIA labels, keyboard support, clipboard API fallback)
- B.2 Syntax Highlighting: 9.25/10 - Excellent implementation with
  500+ language support and graceful error handling

## TASKS.md Updates

- Added detailed grading with justifications
- Added known limitations for each completed task
- Added test coverage gaps to address
- Added technical specifications (file architecture, CSS/JS guidelines)
- Added task dependency graph
- Added implementation details for pending phases
- Added documentation gaps identified

## AGENTS.md Updates

- Added Quick Start section with setup commands
- Added Project Structure overview
- Added comprehensive testing commands
- Added snapshot testing instructions
- Added debugging tips
- Added architecture notes

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

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Author: ShlomoStept

AGENTS.md

@@ -1,15 +1,161 @@
-Uses uv. Run tests like this:
+# Development Guide
 
-    uv run pytest
+This guide covers everything needed to contribute to claude-code-transcripts.
 
-Run the development version of the tool like this:
+## Quick Start
 
-    uv run claude-code-transcripts --help
+```bash
+# Clone and setup
+git clone https://github.com/simonw/claude-code-transcripts.git
+cd claude-code-transcripts
 
-Always practice TDD: write a faliing test, watch it fail, then make it pass.
+# Install uv if not already installed
+# See: https://docs.astral.sh/uv/
 
-Commit early and often. Commits should bundle the test, implementation, and documentation changes together.
+# Install dependencies
+uv sync --group dev
 
-Run Black to format code before you commit:
+# Run tests
+uv run pytest
 
-    uv run black .
+# Run the development version
+uv run claude-code-transcripts --help
+```
+
+## Project Structure
+
+```
+claude-code-transcripts/
+├── src/claude_code_transcripts/
+│   ├── __init__.py          # Main implementation
+│   └── templates/           # Jinja2 templates
+│       ├── macros.html      # Reusable macros
+│       ├── page.html        # Page template
+│       ├── index.html       # Index template
+│       ├── base.html        # Base template
+│       └── search.js        # Client-side search
+├── tests/
+│   ├── test_generate_html.py  # Main test suite
+│   ├── test_all.py            # Batch command tests
+│   ├── sample_session.json    # Test fixture (JSON)
+│   ├── sample_session.jsonl   # Test fixture (JSONL)
+│   └── __snapshots__/         # Snapshot test outputs
+├── TASKS.md                   # Implementation roadmap
+├── AGENTS.md                  # This file
+└── pyproject.toml             # Package configuration
+```
+
+## Running Tests
+
+```bash
+# Run all tests
+uv run pytest
+
+# Run specific test file
+uv run pytest tests/test_generate_html.py
+
+# Run specific test class
+uv run pytest tests/test_generate_html.py::TestRenderContentBlock
+
+# Run specific test
+uv run pytest tests/test_generate_html.py::TestRenderContentBlock::test_text_block -v
+
+# Run with verbose output
+uv run pytest -v
+
+# Run with stdout capture disabled (for debugging)
+uv run pytest -s
+```
+
+## Code Formatting
+
+Format code with Black before committing:
+
+```bash
+uv run black .
+```
+
+Check formatting without making changes:
+
+```bash
+uv run black . --check
+```
+
+## Test-Driven Development (TDD)
+
+Always practice TDD: write a failing test, watch it fail, then make it pass.
+
+1. Write a failing test for your change
+2. Run tests to confirm it fails: `uv run pytest`
+3. Implement the feature to make the test pass
+4. Format your code: `uv run black .`
+5. Run all tests to ensure nothing broke
+6. Commit with a descriptive message
+
+## Snapshot Testing
+
+This project uses `syrupy` for snapshot testing. Snapshots are stored in `tests/__snapshots__/`.
+
+Update snapshots when intentionally changing output:
+
+```bash
+uv run pytest --snapshot-update
+```
+
+## Making Changes
+
+### Commit Guidelines
+
+Commit early and often. Each commit should bundle:
+- The test
+- The implementation
+- Documentation changes (if applicable)
+
+Example commit message:
+```
+Add support for filtering sessions by date
+
+- Add --since and --until flags to local command
+- Filter sessions by modification time
+- Add tests for date filtering
+```
+
+### Before Submitting a PR
+
+1. All tests pass: `uv run pytest`
+2. Code is formatted: `uv run black .`
+3. Documentation updated if adding user-facing features
+4. TASKS.md updated if completing a tracked task
+
+## Key Files Reference
+
+| File | Purpose |
+|------|---------|
+| `src/claude_code_transcripts/__init__.py` | Main implementation (~1300 lines) |
+| `src/claude_code_transcripts/templates/macros.html` | Jinja2 macros for rendering |
+| `tests/test_generate_html.py` | Main test suite |
+| `tests/sample_session.json` | Test fixture data |
+| `TASKS.md` | Implementation roadmap and status |
+
+## Debugging Tips
+
+```bash
+# See full assertion output
+uv run pytest -vv
+
+# Stop on first failure
+uv run pytest -x
+
+# Run only failed tests from last run
+uv run pytest --lf
+
+# Run tests matching a pattern
+uv run pytest -k "test_ansi"
+```
+
+## Architecture Notes
+
+- CSS and JavaScript are embedded as string constants in `__init__.py`
+- Templates use Jinja2 with autoescape enabled
+- The `_macros` module exposes macros from `macros.html`
+- Tool rendering follows the pattern: Python function → Jinja2 macro → HTML