Document a2a

README.md

@@ -18,6 +18,7 @@ Retrieval-Augmented Generation (RAG) library built on LanceDB.
 - **File monitoring**: Auto-index files when run as server
 - **40+ file formats**: PDF, DOCX, HTML, Markdown, code files, URLs
 - **MCP server**: Expose as tools for AI assistants
+- **A2A agent**: Conversational agent with context and multi-turn dialogue
 - **CLI & Python API**: Use from command line or Python
 
 ## Quick Start
@@ -143,6 +144,20 @@ haiku-rag serve --stdio
 
 Provides tools for document management and search directly in your AI assistant.
 
+## A2A Agent
+
+Run as a conversational agent with the Agent-to-Agent protocol:
+
+```bash
+haiku-rag serve --a2a
+```
+
+Provides a conversational interface with:
+- Multi-turn dialogue with context
+- Intelligent multi-search for complex questions
+- Source citations with titles and URIs
+- Full document retrieval on request
+
 ## Documentation
 
 Full documentation at: https://ggozad.github.io/haiku.rag/

docs/a2a.md

@@ -0,0 +1,83 @@
+# Agent-to-Agent (A2A) Protocol
+
+The A2A server exposes `haiku.rag` as a conversational agent using the Agent-to-Agent protocol. Unlike the MCP server which provides stateless tools, the A2A agent maintains conversation history and context across multiple turns.
+
+## Features
+
+- **Conversational Context**: Maintains full conversation history including tool calls and results
+- **Multi-turn Dialogue**: Supports follow-up questions with pronoun resolution ("he", "it", "that document")
+- **Intelligent Search**: Performs single or multiple searches depending on question complexity
+- **Source Citations**: Always includes sources with both titles and URIs
+- **Full Document Retrieval**: Can fetch complete documents on request
+- **Document Discovery**: Lists available documents to help users explore the knowledge base
+
+## Starting A2A Server
+
+```bash
+haiku-rag serve --a2a
+```
+
+Server options:
+- `--a2a-host` - Host to bind to (default: 127.0.0.1)
+- `--a2a-port` - Port to bind to (default: 8000)
+
+Example:
+```bash
+haiku-rag serve --a2a --a2a-host 0.0.0.0 --a2a-port 8080
+```
+
+## Requirements
+
+A2A support requires the `a2a` extra:
+
+```bash
+uv pip install 'haiku.rag[a2a]'
+```
+
+## Python Usage
+
+```python
+from pathlib import Path
+from haiku.rag.a2a import create_a2a_app
+import uvicorn
+
+# Create A2A app
+app = create_a2a_app(Path("database.lancedb"))
+
+# Run with uvicorn
+uvicorn.run(app, host="127.0.0.1", port=8000)
+```
+
+This installs the `fasta2a` package and its dependencies.
+
+## Architecture
+
+The A2A agent uses:
+
+- **FastA2A**: Python framework implementing the A2A protocol
+- **Pydantic AI**: Agent framework with tool support
+- **In-Memory Storage**: Context and message history storage (persists during server lifetime)
+- **Conversation State**: Full pydantic-ai message history serialized in A2A context
+
+### Message History
+
+The agent stores the complete conversation state including:
+
+- User prompts
+- Agent responses
+- Tool calls and their arguments
+- Tool return values
+
+This enables the agent to:
+
+- Reference previous searches
+- Understand pronouns and context
+- Maintain coherent multi-turn conversations
+
+### Context Management
+
+Each conversation is identified by a `context_id`. All messages within the same context share conversation history. This allows the agent to:
+
+- Remember what was discussed
+- Track which documents were already found
+- Provide contextual follow-up answers

docs/index.md

@@ -10,10 +10,11 @@
 - **Support for various embedding providers**: Ollama, VoyageAI, OpenAI or add your own
 - **Native Hybrid Search**: Vector search combined with full-text search using native LanceDB RRF reranking
 - **Reranking**: Optional result reranking with MixedBread AI or Cohere
-- **Question Answering**: Built-in QA agents using Ollama, OpenAI, or Anthropic.
+- **Question Answering**: Built-in QA agents using Ollama, OpenAI, or Anthropic
 - **File monitoring**: Automatically index files when run as a server
 - **Extended file format support**: Parse 40+ file formats including PDF, DOCX, HTML, Markdown, code files and more. Or add a URL!
 - **MCP server**: Exposes functionality as MCP tools
+- **A2A agent**: Conversational agent with context and multi-turn dialogue support
 - **CLI commands**: Access all functionality from your terminal
   - Add sources from text, files, or URLs, optionally with a human‑readable title
 - **Python client**: Call `haiku.rag` from your own python applications
@@ -57,6 +58,7 @@ haiku-rag migrate old_database.sqlite  # Migrate from SQLite
 - [CLI](cli.md) - Command line interface usage
 - [Server](server.md) - File monitoring and server mode
 - [MCP](mcp.md) - Model Context Protocol integration
+- [A2A](a2a.md) - Agent-to-Agent conversational protocol
 - [Python](python.md) - Python API reference
 - [Agents](agents.md) - QA agent and multi-agent research
 

docs/server.md

@@ -1,9 +1,11 @@
 # Server Mode
 
-The server provides automatic file monitoring and MCP functionality.
+The server provides automatic file monitoring, MCP functionality, and A2A agent support.
 
 ## Starting the Server
 
+### MCP Server (Default)
+
 ```bash
 haiku-rag serve
 ```
@@ -12,6 +14,18 @@ Transport options:
 - Default - Streamable HTTP transport
 - `--stdio` - Standard input/output transport
 
+### A2A Server
+
+```bash
+haiku-rag serve --a2a
+```
+
+Options:
+- `--a2a-host` - Host to bind to (default: 127.0.0.1)
+- `--a2a-port` - Port to bind to (default: 8000)
+
+See [A2A documentation](a2a.md) for details on the conversational agent.
+
 ## File Monitoring
 
 Set `MONITOR_DIRECTORIES` environment variable to enable automatic file monitoring:

mkdocs.yml

@@ -64,6 +64,7 @@ nav:
       - Agents: agents.md
       - Python: python.md
       - MCP: mcp.md
+      - A2A: a2a.md
       - Benchmarks: benchmarks.md
 markdown_extensions:
   - admonition