{"id":"cdatasoftware-amazon-redshift-mcp-server-by-cdata","name":"amazon-redshift-mcp-server-by-cdata","homepage":null,"repo_url":"https://github.com/CDataSoftware/amazon-redshift-mcp-server-by-cdata","category":"ai-ml","subcategories":[],"tags":["mcp","amazon-redshift","jdbc","data-access","llm-integration","local-stdio"],"what_it_does":"Provides a local MCP (Model Context Protocol) server that wraps the CData JDBC Driver for Amazon Redshift, exposing database metadata (tables/columns) and query execution via MCP tools so an LLM client can retrieve Redshift data (README describes read-only as the project’s intent).","use_cases":["Use an AI desktop client (e.g., Claude Desktop) to query live Amazon Redshift data without writing SQL","Table/column discovery for Redshift schemas via MCP tools","Ad-hoc read-only analytics questions against Redshift using the MCP tool interface"],"not_for":["Public/hosted multi-tenant deployment (README indicates stdio/local-only usage)","Environments that require robust server-side auth, rate limiting, or audit logging from this MCP server itself","Use cases requiring guaranteed safety against unintended broad SQL queries (client issues run_query tool)"],"best_when":"You want a local, developer-controlled MCP integration to let an LLM explore and run SELECT queries against Redshift over stdio on the same machine.","avoid_when":"You need network-accessible API endpoints, fine-grained API authorization managed by the MCP server, or you cannot control what SQL the LLM/client may generate.","alternatives":["CData-managed MCP platform (CData Connect AI / managed MCP offering referenced in README)","Direct access to Amazon Redshift via a backend API you control (REST/GraphQL) and run LLM-to-SQL through your service","Other MCP servers or database agent frameworks that provide schema discovery and safe query execution (where available)"],"af_score":50.0,"security_score":49.8,"reliability_score":18.8,"package_type":"mcp_server","discovery_source":["github"],"priority":"low","status":"evaluated","version_evaluated":null,"last_evaluated":"2026-04-04T19:49:30.306623+00:00","interface":{"has_rest_api":false,"has_graphql":false,"has_grpc":false,"has_mcp_server":true,"mcp_server_url":null,"has_sdk":false,"sdk_languages":[],"openapi_spec_url":null,"webhooks":false},"auth":{"methods":["CData JDBC driver authentication mechanisms (including OAuth when specified by JdbcUrl/connection string)"],"oauth":true,"scopes":false,"notes":"The README describes that OAuth may be required depending on the data source and that the JDBC connection string (e.g., JdbcUrl with InitiateOAuth=GETANDREFRESH) is placed into a local .prp config used by the MCP server. MCP server-level scopes are not described."},"pricing":{"model":null,"free_tier_exists":false,"free_tier_limits":null,"paid_tiers":["Requires CData JDBC Driver licensing (driver license step in README).","Project itself is MIT-licensed; pricing depends on the CData driver/managed offerings referenced."],"requires_credit_card":false,"estimated_workload_costs":null,"notes":"No pricing details for the MCP server are provided in the README; the driver license is required."},"requirements":{"requires_signup":false,"requires_credit_card":false,"domain_verification":false,"data_residency":[],"compliance":[],"min_contract":null},"agent_readiness":{"af_score":50.0,"security_score":49.8,"reliability_score":18.8,"mcp_server_quality":78.0,"documentation_accuracy":70.0,"error_message_quality":0.0,"error_message_notes":null,"auth_complexity":55.0,"rate_limit_clarity":0.0,"tls_enforcement":80.0,"auth_strength":60.0,"scope_granularity":20.0,"dependency_hygiene":45.0,"secret_handling":40.0,"security_notes":"Strengths: The connection to Redshift is via the CData JDBC driver and the setup mentions OAuth as an option, suggesting use of established auth flows at the driver layer. Weaknesses/unknowns: The MCP server itself is local-stdio and does not describe TLS termination (transport is stdio, not a network API), server-side auth, request authorization, or audit logging. Scope granularity is not described (no fine-grained MCP scopes). Secrets/connection strings are stored in a local .prp file (plaintext on disk); the README does not describe secure secret handling or redaction. SQL is executed through a generic run_query tool; unless the driver/MCP layer restricts non-SELECT, there is risk from prompt-generated SQL. Dependency hygiene and vulnerability posture are not evidenced in the provided content.","uptime_documented":0.0,"version_stability":45.0,"breaking_changes_history":0.0,"error_recovery":30.0,"idempotency_support":"false","idempotency_notes":"Tools include run_query (SELECT) and metadata reads; idempotency depends on the SQL executed. The README frames the server as read-only but does not specify idempotent behavior or safeguards (e.g., forbidding non-SELECT).","pagination_style":"none","retry_guidance_documented":false,"known_agent_gotchas":["Server uses stdio and therefore only works with clients running on the same machine (per README).","The tools allow SQL SELECT execution via a run_query tool; LLMs may generate overly broad or expensive queries unless constrained.","The README does not document pagination/limits; large result sets may be truncated or expensive depending on driver/client behavior.","Tool outputs are described as CSV for tables/columns; agents should handle CSV parsing correctly."]}}