Multi-account AWS security scanning and infrastructure ledger with an MCP server for querying scan history
CloudLedger provides 11 well-named, read-only AWS security tools with mostly complete schemas and descriptions. Naming follows verb_noun convention (list_, get_, find_, search_, compare_). Descriptions are present and contextual (avg ~150 chars), exceeding the 10-char minimum. All tools have input schemas with typed properties. However, output schemas are entirely undocumented, LLMs cannot predict response structure for chaining. Error handling is absent; no recovery guidance or actionable error messages. Parameters lack validation constraints (e.g., search_by_ip accepts both ip_address and cidr_range but doesn't clarify mutual exclusivity). Some descriptions are generic ('analyse their rules' for find_security_group_rules). No tool annotations (readOnlyHint, idempotentHint) despite all being read-only. Missing pagination guidance for tools returning large result sets (list_scans, get_iam_users).
Compare resource counts between two scans to identify changes
Find EC2 instances filtered by IP address type: public (has public IP), private (no public IP, only private), or all instances. Supports pagination for large result sets and summary mode for aggregated counts.
Find S3 buckets that may be publicly accessible
Find security groups and analyse their rules
Get IAM users with optional filtering
Get load balancers (ALB/NLB/Classic) for a scan
Get Prowler security findings with comprehensive summarisation and optional filtering. Prowler performs automated security assessments against AWS best practices, compliance frameworks (CIS, PCI-DSS, GDPR, HIPAA, etc.), and security standards. Returns detailed findings with an always-included summary containing: severity breakdown (critical/high/medium/low/informational), status distribution, top failing services, top failing check types, compliance framework coverage, and actionable insights (e.g., "10 CRITICAL findings require immediate attention"). Use summary_mode=true to get only the summary without detailed findings.
Output schemas completely undocumented. LLMs cannot predict response structure, field names, or types. Breaks tool chaining and forces agents to guess what data is available.
No error handling or recovery guidance. Tools provide no actionable error messages, categorization (retryable vs fatal), or suggestions for next steps when calls fail.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | <=2025-11-25 | v2 |
Get summary of resources collected in a specific scan
Get detailed VPC architecture including subnets, security groups, and instances
List all AWS scans or scans for a specific account
Search for AWS resources by IP address or CIDR range
search_by_ip accepts both 'ip_address' and 'cidr_range' parameters but does not document mutual exclusivity or which takes precedence. LLMs may pass both, causing ambiguous behavior.
No tool annotations (readOnlyHint, idempotentHint, destructiveHint). All 11 tools are read-only and idempotent, but LLMs cannot infer this without explicit hints.
Pagination not documented for tools returning large result sets. list_scans, get_iam_users, and others lack guidance on limit/offset behavior, total counts, or when to paginate.