A Go-based runtime platform for orchestrating multi-agent workflows with provider connectors, activity tools, and semantic contract enforcement
The server exposes only one tool (read_flow_data) with a minimal schema. The tool definition appears in a test file (tier12_runtime_tools_e2e_test.go), not in production code, suggesting it is inferred rather than explicitly registered. The description is present but generic (58 chars). The input schema has two parameters (kind, static_id) with type strings and basic descriptions, but lacks critical metadata: no enums for 'kind', no format/pattern constraints, no output schema documentation. The codebase shows a complex Go application with provider connectors, packs, and runtime orchestration, but no visible MCP server implementation, tool registration mechanism, or protocol handler. This appears to be a non-MCP application that was incorrectly classified as an MCP server.
Reads static flow data files with access control based on declared flow data access permissions
Tool definition inferred from test file, not explicitly registered in production code. No visible MCP server implementation, tool registration API, or protocol handler detected.
Input schema lacks enums for 'kind' parameter. Free-form string invites hallucinated values like 'dynamic_file' or 'api_endpoint' instead of valid options.
No output schema documented. LLMs cannot plan downstream tool calls or extract required fields without knowing what read_flow_data returns.
Description (58 chars) is below baseline (194 chars avg). Does not explain WHEN to use this tool, what prerequisites exist, or how it differs from other data access patterns.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 31 | 2025-06-18+ | v2 |
No error handling guidance. If 'static_id' is not found or access is denied, what does the tool return? How should the LLM recover?