Restructure docs and clean up homepage

- Remove emojis from homepage headings, nav, and buttons
- Fix How It Works section: remove extra margin, add connecting
  lines between steps, constrain paragraph width
- Add "Read the Docs" CTA button to hero and footer
- Restructure docs into new flow: Quick Start, Integrations,
  Custom Evaluators, UI Walkthrough, Advanced, FAQ
- Consolidate old pages (ci-cd, configuration, examples,
  mcp-server, web-ui) into new Integrations page
- Add Custom Evaluators page with matching modes, SDK examples,
  and link to repo guide
- Add FAQ page with common questions
- Add Advanced page with config reference and API tables

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: 2 people

content/docs/_index.md

@@ -1,3 +1,4 @@
 ---
 title: "Documentation"
+description: "Everything you need to get started with AgentEvals."
 ---

content/docs/advanced.md

@@ -0,0 +1,123 @@
+---
+title: "Advanced"
+weight: 5
+description: "Advanced configuration, API reference, and deep-dive resources."
+---
+
+## Configuration Reference
+
+### Eval Set Configuration
+
+```yaml
+# agentevals.yaml
+version: "1"
+
+trace_sources:
+  - type: otlp
+    port: 4318
+    protocol: http
+
+  - type: jaeger
+    path: ./traces/*.json
+
+llm:
+  provider: openai
+  model: gpt-4o
+  temperature: 0.0
+
+output:
+  format: table    # table, json, junit
+  verbose: false
+```
+
+### Evaluator Parameters
+
+| Parameter | Type | Default | Description |
+|-----------|------|---------|-------------|
+| `trajectory_match_mode` | `"strict"` \| `"unordered"` \| `"subset"` \| `"superset"` | `"strict"` | How to compare trajectories |
+| `tool_args_match_mode` | `"exact"` \| `"ignore"` \| `"subset"` \| `"superset"` | `"exact"` | How to match tool arguments |
+| `tool_args_match_overrides` | `Dict[str, ...]` | `None` | Custom matchers per tool |
+| `model` | `str` | `None` | LLM model for judge evaluators |
+| `continuous` | `bool` | `False` | Float (0–1) vs boolean scoring |
+| `use_reasoning` | `bool` | `True` | Include reasoning in output |
+| `few_shot_examples` | `List[FewShotExample]` | `None` | Example evaluations for LLM judge |
+| `feedback_key` | `str` | `"trajectory_accuracy"` | Key name for evaluation results |
+
+### Environment Variables
+
+| Variable | Description | Default |
+|----------|-------------|---------|
+| `AGENTEVALS_CONFIG` | Path to config file | `./agentevals.yaml` |
+| `OPENAI_API_KEY` | OpenAI API key for LLM judge | — |
+| `ANTHROPIC_API_KEY` | Anthropic API key for LLM judge | — |
+| `AGENTEVALS_LOG_LEVEL` | Log level (debug, info, warn, error) | `info` |
+| `AGENTEVALS_OUTPUT_FORMAT` | Output format override | `table` |
+
+## Deep-Dive Documentation
+
+For comprehensive coverage of specific topics, see the repository docs:
+
+- [Trajectory Match Evaluators](https://github.com/agentevals-dev/agentevals#trajectory-match-evaluators) — Full reference for all matching modes and configuration
+- [LLM-as-Judge Evaluators](https://github.com/agentevals-dev/agentevals#llm-as-judge-evaluators) — Custom prompts, few-shot examples, continuous scoring
+- [Graph Trajectory Evaluators](https://github.com/agentevals-dev/agentevals#graph-trajectory-evaluators) — LangGraph integration and trajectory extraction
+- [LangSmith Integration](https://github.com/agentevals-dev/agentevals#langsmith-integration) — Running evaluations with pytest, Vitest, and LangSmith experiments
+- [Custom Evaluators Guide](https://github.com/agentevals-dev/agentevals/blob/main/docs/custom-evaluators.md) — Writing domain-specific evaluators
+
+## Async Support
+
+Both Python and TypeScript support fully async evaluators:
+
+### Python
+
+```python
+from agentevals.trajectory.match import (
+    create_async_trajectory_match_evaluator
+)
+from agentevals.trajectory.llm import (
+    create_async_trajectory_llm_as_judge
+)
+
+# Async trajectory match
+async_match = create_async_trajectory_match_evaluator(
+    trajectory_match_mode="strict"
+)
+result = await async_match(
+    outputs=trajectory,
+    reference_outputs=reference
+)
+
+# Async LLM-as-judge
+async_judge = create_async_trajectory_llm_as_judge(
+    model="openai:o3-mini"
+)
+result = await async_judge(
+    outputs=trajectory,
+    reference_outputs=reference
+)
+```
+
+## API Reference
+
+### Python Public API
+
+| Function | Description |
+|----------|-------------|
+| `create_trajectory_match_evaluator()` | Create a configurable trajectory match evaluator |
+| `create_async_trajectory_match_evaluator()` | Async variant |
+| `create_trajectory_llm_as_judge()` | Evaluate trajectory quality with an LLM judge |
+| `create_async_trajectory_llm_as_judge()` | Async variant |
+| `create_graph_trajectory_llm_as_judge()` | LLM-as-judge for LangGraph workflows |
+| `create_async_graph_trajectory_llm_as_judge()` | Async variant |
+| `graph_trajectory_strict_match()` | Strict match for graph execution steps |
+| `extract_langgraph_trajectory_from_thread()` | Extract trajectory from a LangGraph thread |
+| `extract_langgraph_trajectory_from_snapshots()` | Extract trajectory from state snapshots |
+
+### TypeScript Public API
+
+| Function | Description |
+|----------|-------------|
+| `createTrajectoryMatchEvaluator()` | Create a configurable trajectory match evaluator |
+| `createTrajectoryLLMAsJudge()` | Evaluate trajectory quality with an LLM judge |
+| `createGraphTrajectoryLLMAsJudge()` | LLM-as-judge for LangGraph workflows |
+| `extractLangGraphTrajectoryFromThread()` | Extract trajectory from a LangGraph thread |
+| `extractLangGraphTrajectoryFromSnapshots()` | Extract trajectory from state snapshots |