diff --git a/README.md b/README.md index fbdb6f9..16f936d 100644 --- a/README.md +++ b/README.md @@ -28,13 +28,13 @@ make install make test ``` -Run an agent: +Run a specialized agent for codebase exploration: ```bash -uv --project agentic-framework run agentic-run simple -i "Explain what an MCP server is." +uv --project agentic-framework run agentic-run developer -i "List the project structure and explain what the main components do." ``` -List available agents: +List all available agents: ```bash uv --project agentic-framework run agentic-run list @@ -49,8 +49,8 @@ Run the framework in Docker with mounted volumes for live code updates: make docker-build # Run agents using the shell wrapper +bin/agent.sh developer -i "Search for all Tool definitions in the project" bin/agent.sh chef -i "I have bread, tuna, lettuce and mayo" -bin/agent.sh -v travel-coordinator -i "Plan a trip to Paris" bin/agent.sh list # Access logs (same location as local) @@ -65,17 +65,28 @@ tail -f agentic-framework/logs/agent.log - ✅ Simple shell wrapper mimics local CLI experience -## Current Agents +## Featured Agents | Agent | What it does | MCP Access | |---|---|---| -| `simple` | basic conversational chain | none | -| `chef` | recipe suggestions from ingredients | `tavily` | -| `travel` | flight planning assistant | `kiwi-com-flight-search` | -| `news` | AI news assistant | `web-fetch` | -| `travel-coordinator` | orchestrates 3 specialists (flights + city intel + reviewer) | `kiwi-com-flight-search`, `web-fetch` | +| `developer` | 🚀 **Principal Engineer** assistant for codebase exploration and search | `web-fetch` | +| `travel-coordinator` | ✈️ **Multi-agent Orchestrator** (Flights + City Intel + Reviewer) | `kiwi-com-flight-search`, `web-fetch` | +| `chef` | 🍳 Recipe suggestions from ingredients | `tavily` | +| `news` | 📰 AI news assistant (TechCrunch specialist) | `web-fetch` | +| `travel` | 🎫 Flight planning assistant | `kiwi-com-flight-search` | +| `simple` | 💬 Basic conversational chain | none | -## Coordinator Example (3 Agents + 2 MCP Servers) +## Showcase: Developer Agent + +The `developer` agent is designed to assist with codebase maintenance and understanding. It comes equipped with local tools for: +- **File Discovery**: Finding files by name across the project. +- **Structure Exploration**: Visualizing the project directory tree. +- **Code Outlining**: Extracting functions and classes from Python files. +- **Pattern Search**: Global search using `ripgrep` for fast pattern matching. + +Implementation: `src/agentic_framework/core/developer_agent.py` + +## Showcase: Travel Coordinator (Multi-Agent System) Run: @@ -83,24 +94,15 @@ Run: uv --project agentic-framework run agentic-run travel-coordinator -i "Plan a 5-day trip from Lisbon to Berlin in May." ``` -This example uses two remote MCP servers that are publicly configured in this project: -`kiwi-com-flight-search` and `web-fetch`. - -How it works: +This example demonstrates complex orchestration using two remote MCP servers: `kiwi-com-flight-search` and `web-fetch`. -1. `FlightSpecialistAgent` uses MCP tools to gather flight options. -2. `CityIntelAgent` receives the flight report and adds destination intelligence. -3. `TravelReviewerAgent` receives both reports and returns the final itinerary brief. +**How it works:** -The coordinator implementation lives in: +1. **FlightSpecialistAgent** uses MCP tools to gather flight options. +2. **CityIntelAgent** receives the flight report and adds destination intelligence. +3. **TravelReviewerAgent** receives both reports and returns the final itinerary brief. -- `agentic-framework/src/agentic_framework/core/travel_coordinator_agent.py` - -Why this matters: - -- You can build orchestration logic in a new agent file. -- You can compose many specialist agents without touching the framework base. -- You can choose per-agent MCP access in registry metadata only. +The coordinator implementation lives in: `src/agentic_framework/core/travel_coordinator_agent.py` ## Architecture (Beginner-Friendly) @@ -160,6 +162,7 @@ After adding the file under `src/agentic_framework/core/`, the CLI command appea uv --project agentic-framework run agentic-run my-agent -i "hello" ``` + ## Scaling to Coordinators and Multi-Agent Systems Recommended approach: