A Go-based framework for building AI agents with tool calling, memory management, and provider abstraction for multiple LLM providers (OpenAI, Anthropic, Google, Bedrock)
Galdor is a Go framework/library for building agentic systems, not an MCP server. The single tool 'add' is defined in a test file (pkg/observability/tool_test.go), suggesting it is a test fixture rather than a production tool. The tool has a minimal description ('Add two integers') and a basic schema with two integer parameters. No evidence of MCP protocol implementation, transport layer, or server capability. The codebase appears to be an agent framework with OpenTelemetry integration, not an MCP-compliant tool server.
Add two integers
Tool defined in test file, not production code. Unclear if 'add' is a real tool or a test fixture.
Description is extremely brief (18 chars). Does not explain when to use the tool, what it returns, or any prerequisites. Fails pattern:tool-description baseline of 34-392 chars.
No output schema documented. LLMs cannot infer what the tool returns or plan downstream calls.
No MCP protocol implementation detected. No evidence of transport layer (HTTP, SSE, STDIO), tool registration, or MCP message handling.
No error handling or recovery guidance. Tool provides no context for LLM to handle failures (e.g., integer overflow, invalid input).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 32 | 2026-07-28+ | v2 |