MCP server for interacting with ReportPortal, a test automation result management platform. Supports launching tests, managing test items, and TMS integration.
The ReportPortal MCP server has a single tool 'get_launches' with minimal documentation. The tool is referenced in server_test.go but complete tool registration code is not visible in the provided source. Input schema shows only a 'projectKey' parameter with basic type and description, but output schema is completely undocumented. No parameter validation rules, constraints, enums, or error handling guidance are evident. The tool description is generic (23 characters: 'Retrieve launches from ReportPortal') and lacks context about when to use it, what it returns, pagination support, or performance characteristics. No evidence of tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite the tool being marked READ_ONLY in metadata. The codebase shows sophisticated HTTP server infrastructure (Chi router, TLS, connection pooling, analytics) but the MCP tool layer itself is underdeveloped.
Retrieve launches from ReportPortal
Tool description is too generic and lacks LLM guidance. 'Retrieve launches from ReportPortal' (23 chars) does not explain when to use this tool, what a 'launch' is in context, pagination behavior, or expected response structure.
Output schema completely undocumented. No visible specification of return type, structure, fields, or pagination. LLMs cannot plan downstream actions or extract needed data without knowing what fields to expect.
No pagination support specified despite tool name ('get_launches' plural suggests multiple results). Missing page/offset/limit parameters and total count or next_cursor in response. Large result sets risk context window exhaustion.
Parameter 'projectKey' has minimal description ('The ReportPortal project key') but lacks format constraints. No specification of valid length (2-10 chars?), character set (alphanumeric?), or case sensitivity. LLMs will guess at valid values.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 43 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 13 | - | v1 |
Tool registration inferred from test file reference. Actual tool definition code (tool registration with schema, handler attachment) not visible in provided source. Cannot verify complete schema structure, constraints, or error handling.
No error handling guidance provided. No documentation of what errors can occur (invalid projectKey, authentication failure, API timeout), how to classify them (retryable vs user-fixable), or recovery steps.
Tool marked READ_ONLY in metadata but no readOnlyHint annotation present in schema. Current spec requires tool annotations (readOnlyHint, destructiveHint, idempotentHint) for LLM optimization of planning and caching.
Single tool with no composition or chaining. No indication of what reference IDs (launch_id, etc.) are returned that might be needed for follow-up operations. Limits agent ability to compose workflows.
Credentials and authentication token handling not visible in tool parameters (good, indicates server-side injection), but no documentation of how authentication is enforced or what happens on auth failure.