An MCP server that provides tools for accessing AWS CloudWatch logs.
Server has 2 tools with partial schema coverage and basic descriptions, but suffers from critical issues: (1) Credentials exposed as parameters (security anti-pattern), (2) descriptions under recommended length, (3) minimal error guidance, (4) output schemas not documented. Tool naming is acceptable (verb_noun pattern), but parameter design violates secret-injection pattern. Both tools are defined and visible in main.py with schemas, placing them above complete absence, but multiple high-severity gaps prevent a higher score.
Get CloudWatch logs from a specific log group and stream
List available CloudWatch log groups
AWS credentials (accessKeyId, secretAccessKey, sessionToken) exposed as tool parameters
Output schemas not documented, LLM cannot reason about response structure
No result limits or pagination, get_logs can return unbounded event lists
Descriptions under recommended length, insufficient LLM optimization
Time format parsing not formalized, description mentions 'ISO format or relative format' but no constraint
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 44 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 42 | - | v1 |
Error handling returns JSON error object instead of actionable guidance