Agent Executor (AX) is Google's open agentic orchestration runtime
Go
Simplify the README
JBD68d4e132fe49
diff --git a/README.md b/README.mdindex c957f77..74d839c 100644--- a/README.md+++ b/README.md@@ -74,7 +74,7 @@ The remote agent runs as a gRPC server implementing AgentService on port :50051.**Terminal 2** - Start the gar controller server:```bash-gar serve --addr :8494+gar serve```The gar server exposes the GARService on port :8494.@@ -95,9 +95,6 @@ gar trigger \--session-id session123 \--input "Hello remote agent"-# List all sessions-gar list --server localhost:8494-# Inspect session detailsgar inspect --server localhost:8494 --session-id session123```@@ -187,7 +184,7 @@ Options:- `--name`: Human-readable name for the agent- `--description`: Description of agent capabilities-#### Run Controller Server+#### Run Server```bashgar serve [--config <path>]@@ -213,7 +210,6 @@ eventlog:dir: "eventlog"controller:- max_steps: 100health_check_interval: 30s```@@ -226,48 +222,6 @@ gar servegar serve --config my-config.yaml```-Once running, clients can connect to the server to:-- Start and resume sessions remotely-- Query session status and list sessions-- Register and unregister agents dynamically--### Gemini-Powered Agent Selection--GAR uses Google's Gemini models by default for intelligent agent selection. Instead of simple round-robin routing, Gemini analyzes the conversation context and agent capabilities to choose the best agent for each task.--**Setup:**--1. Get an API key from [Google AI Studio](https://aistudio.google.com/app/apikey)-2. Set the environment variable:- ```bash- export GEMINI_API_KEY="your-api-key-here"- ```-3. Start the server:- ```bash- gar serve- ```--**How it works:**-- Each registered agent becomes a "tool" that Gemini can call-- Gemini reviews conversation history and agent descriptions-- Gemini selects the most appropriate agent for each step-- Agent metadata helps Gemini make better routing decisions--**Environment Variables:**-- `GEMINI_API_KEY`: Your Google AI API key (required)-- `GAR_GEMINI_MODEL`: Model name (optional, defaults to "gemini-flash-latest")--**Example agent registration with metadata:**-```bash-gar register \- --server localhost:8494 \- --agent-id math-agent \- --name "Math Agent" \- --description "Specialized in mathematical calculations and problem solving"-```--When a user asks "What is 25 * 4?", Gemini will automatically route to the math-agent instead of a generic agent.-### CheckpointsCheckpoints provide a mechanism to save and resume session state at specific points. Every content event (both `CONTENT_IN` and `CONTENT_OUT`) automatically creates a checkpoint with a unique UUID.@@ -405,7 +359,7 @@ func (s *server) HealthCheck(ctx context.Context, req *proto.HealthCheckRequest)**Workflow:**1. Remote agent starts as gRPC server on a port (e.g., :50051)-2. Start gar controller: `gar serve --addr :8494`+2. Start gar controller: `gar serve`3. Register with gar: `gar register --server localhost:8494 --agent-id my-agent --name "My Agent" --description "Agent description" --agent-addr localhost:50051`4. When gar triggers a session, it calls the agent's `Process` RPC5. GAR streams input content → Agent processes → Agent streams output back