Skip to main content
Glama
ROADMAP.md1.96 kB
# Implementation Roadmap ## Status Overview **Current Phase**: Phase 2/3 (Integration & Agents) **Overall Status**: ✅ Core Executor Working, Tool Discovery Implemented ## Completed Phases ### ✅ Phase 0: Ground-Truth Setup - **Goal**: Verify APIs and create minimal scaffold. - **Delivered**: - `mcp_servers.json` configuration - `MCPServerManager` for server connection - Tool discovery logic - Sandbox verification ### ✅ Phase 1: Minimal Executor Server - **Goal**: Expose `execute_code` tool. - **Delivered**: - `executor_server.py` using FastMCP - `executor.py` wrapping DSpy's interpreter - Timeout and error handling - Structured result payloads ### ✅ Phase 2: Tool Integration - **Goal**: Make MCP tools available to the LLM. - **Delivered**: - `ToolSchemaFormatter` for readable docs - `mcp_integration.py` helpers - Integration tests for discovery ## Active & Upcoming Phases ### 🚧 Phase 3: Code Generation Agent - **Goal**: Build the DSpy agent that writes the code. - **Tasks**: - Update `agent.py` with `CodeGenerationSignature`. - Implement `available_tools` input field. - Connect `ProgramOfThought` / `CodeAct`. - Create end-to-end examples. ### 📅 Phase 4: Concurrency & Security - **Goal**: Harden the system. - **Tasks**: - Concurrency testing. - Security hardening (imports, network restrictions). - Load testing. ### 📅 Phase 5: Comprehensive Testing - **Goal**: Regression and robustness. - **Tasks**: - Unit tests (ongoing). - Integration tests with real MCP tools. - Property-based tests. ### 📅 Phase 6: Documentation & Demos - **Goal**: Make it usable for others. - **Tasks**: - Jupyter notebooks. - CLI examples. - Finalize docs (You are here). ## Metrics & Success Criteria - **Code Execution**: Success rate >90% - **Latency**: Average <2s - **Security**: 100% timeout enforcement, no unauthorized file access.

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/draphonix/mcp-code-mode'

If you have feedback or need assistance with the MCP directory API, please join our Discord server