suggest_services

Map a use case description to a curated list of AWS services with rationale.

Signature

suggest_services(use_case: str, include_baseline: bool = True) -> dict

Parameters

Name	Type	Default	Description
use_case	str	—	Free-form description of what is being built.
include_baseline	bool	True	Append baseline services (IAM, CloudWatch, CloudTrail).

Returns

{
  "services": [
    { "service": "Amazon Bedrock", "category": "AI/ML", "rationale": "..." }
  ],
  "matched_patterns": ["rag, retrieval"],
  "note": "Matched 1 pattern(s)."
}

Example

"Build a RAG knowledge base over internal product docs."

{
  "services": [
    { "service": "Amazon Bedrock", "category": "AI/ML", "rationale": "Managed access to foundation models (Claude, Nova, Titan) and Knowledge Bases." },
    { "service": "Amazon OpenSearch Serverless (vector)", "category": "Search", "rationale": "Managed vector store with hybrid search; pairs with Bedrock KB out of the box." },
    { "service": "Amazon S3", "category": "Storage", "rationale": "Source of truth for documents fed into the KB." },
    { "service": "AWS Lambda", "category": "Compute", "rationale": "Ingestion and retrieval orchestration." },
    { "service": "AWS IAM", "category": "Security", "rationale": "Least-privilege access — required for every workload." },
    { "service": "Amazon CloudWatch", "category": "Observability", "rationale": "Metrics, logs and alarms for the stack." },
    { "service": "AWS CloudTrail", "category": "Governance", "rationale": "Audit trail of API calls — non-negotiable for any production account." }
  ],
  "matched_patterns": ["rag, retrieval"],
  "note": "Matched 1 pattern(s)."
}

Notes

• Matching is keyword-based and deterministic. See src/mcp_aws_sa/data/service_catalog.py for the full catalog.
• Unmatched use cases still return the baseline services with a hint to refine the description.
• The catalog is intentionally compact — extending it is a one-file diff.