Neon MCP Server vs PlanetScale MCP Server

Neon MCP Server 40

No x402, MPP or L402 (0). Per-unit prices are public, compute per CU-hour and storage per GB-month (20). Free plan with no card (20). A person signs up and authorises in a browser; the CLI can mint a key only after a login (0). 0 + 20 + 20 + 0 = 40.

PlanetScale MCP Server 15

No x402, MPP or L402 in the MCP docs, the pricing page or the source (0). Monthly prices for every cluster size are public without a login, as a Markdown page and as a read-only SQL catalogue, from $5 a month for a single-node Postgres PS-5. Storage, backup and egress rates were not on the page read, and there is no per-call price (15). No free tier or trial was found, and the billing docs say all plans need a valid payment method on file (0). A person signs up in a browser and authorises OAuth or creates a service token in organisation settings (0). Total 15.

Neon MCP Server 90

Every tool has a JSON Schema, many generated from the Neon OpenAPI spec (25). neon.com serves llms.txt and Markdown pages (10). Descriptions say when and when not ("branch_id is a branch id, not a branch name. Call list_branches to resolve a name", "NEVER run autonomously") (18). Typed inputs with enums such as access read_only or read_write, and path arguments checked for . and .. since 2.0.0 (13). Examples in the README and docs, and API 4xx errors classified by kind (11). CHANGELOG with dated releases and a registry version (13). 25 + 10 + 18 + 13 + 11 + 13 = 90.

PlanetScale MCP Server 82

The 14 tools in planetscale/mcp-server have typed zod input schemas, and the other 12 are generated from the public OpenAPI file at planetscale.com/docs/openapi.yaml, 304 operations (25). llms.txt and a Markdown twin of every docs page (10). The custom descriptions state purpose, limits and when to ask the person, for example the 50-second limit, row-level security on Postgres reads and the Checkout sequence for the payment tools. The generated tools are one-line summaries (17). Enums for log levels and summary fields, clamped limits and required fields. organization, database and branch are free strings and SQL is free text by nature (12). The docs give example prompts and a permissions table per tool. Tool errors are plain text beginning Error: with the API status code (10). The product changelog is dated and carries MCP entries, but the server has no version history of its own. The package is 0.0.0, the repository has no tags or changelog, and the registry entry has stayed at 1.0.0 since 15 April 2026 (8). Total 82. The tool definitions were read from the source and the docs, not from the live server, which answered 401 without a token on 9 October.

Neon MCP Server 69

OAuth 2.1 with read and write scopes, per-project and per-role fixed grants, Client ID Metadata Documents and RFC 9207 iss on redirects, or API keys (28). Read-only mode keeps SQL in READ ONLY transactions and withholds connection strings, roleName limits SQL to a role's grants, and destructive tools ask the model to confirm with the user; with an API key anyone who can edit the URL can widen the grant (17). The README warns to review every action and keep the server away from production, but we found no prompt-injection guidance for query results (5). Neon has project audit logs, but we didn't read whether MCP calls appear (5). SOC 2 Type II with a public SOC 3, ISO 27001 and 27701, HIPAA; no security.txt on neon.com (14). 28 + 17 + 5 + 5 + 14 = 69.

PlanetScale MCP Server 77

OAuth with each client registered as an OAuth application, scopes that set no, read-only or full access per organisation or per database, a separate payment-method scope, and service tokens with per-permission grants for headless use. Each query runs on a database credential created for the call and deleted after it (30). Read queries use a reader password on Vitess and a pg_read_all_data role on Postgres that does not bypass row-level security. An insights-only endpoint has no SQL tools, and since 8 October 2026 administrators can cap members' grants at none, read-only or read and write. TRUNCATE and UPDATE or DELETE without WHERE are blocked. DELETE and DDL need confirm_destructive, which the model sets itself, and the check is a regular expression on the start of the statement. Write queries run as the postgres role or a Vitess admin password. Read and write is the default (16). Query results return database content to the model. The docs warn about write access to production, and no prompt-injection guidance or result wrapping was found (4). Organisation audit log and security log on all plans, service tokens showing the IP address they were last used from since 26 October, every MCP query tagged source=planetscale-mcp in Insights, and a list of installed OAuth apps with revoke (13). Vulnerability disclosure policy with security@planetscale.com, SOC 1 Type 2 and SOC 2 Type 2 with HIPAA, PCI DSS for PlanetScale Managed. security.txt returned 404 on 1 November and no bug bounty was found (14). Total 77.

Neon MCP Server 72

Hosted reading. neonstatus.com on status.io with components per region, though no MCP component (20). The page shows only the last 30 days, three incidents, a partial disruption of monitoring data on 21 October (1 hour 22 minutes), slow compute starts in us-east-2 on 2 November (2 hours 14 minutes) and 26 minutes of console and API errors on 2 November; August and September weren't readable (15). The Neon API publishes 700 requests a minute with a 40-a-second burst per route (15). 429 with exponential backoff advice and 423 Locked for overlapping operations with retry guidance, but no idempotency keys (12). An uptime SLA on Scale only, with no percentage on the pricing page (5). The docs give the server no status label and the README says it isn't recommended for production (5). 20 + 15 + 15 + 12 + 5 + 5 = 72.

PlanetScale MCP Server 78

Graded as a hosted service. Status page at planetscalestatus.com on incident.io with per-region components (20). Its incident feed, readable on 1 November, lists six incidents since 3 August, each partial. Postgres management pages and API endpoints returned 500 errors from 12:17 to 12:24 UTC on 2 September, Vitess databases in four GCP regions lost connectivity on 5 September, branch creation was delayed in GCP europe-west-1 on 28 September, AWS us-east-1 cross-zone networking failed on 9 October, us-east-1 dashboard metrics had a gap on 20 October, and on 26 October "certain branches" of Vitess lost connectivity and credential changes were delayed for about 1 hour 40 minutes, which touches the query tools' per-call credentials. Minor only (20). The API reference's rate-limit page gives 600 requests a minute for the PlanetScale API, "subject to change". The MCP docs don't say whether tool calls count against it. Twelve of the tools are generated from the API's own operations, and supabase-mcp, cloudflare-mcp, github-mcp-server and stripe-mcp take their vendor API's published limit for the server (15). The OpenAPI file and the rate-limit page list a 429 response with no Retry-After or backoff guidance, and no idempotency key for writes. The 50-second query limit is documented (3). The SLA commits to 99.99 per cent for single-region and 99.999 per cent for multi-region clusters with service credits, and excludes single-node clusters and beta functions (10). The docs give the MCP server no beta or preview label and the registry entry is 1.0.0 (10). 20 + 20 + 15 + 3 + 10 + 10 = 78.

Neon MCP Server 79

138 tools with no grant, far over 30 (5), with category, projectId and readonly to cut the list (plus 10). List tools take limits and cursors from the API, and describe_branch returns a tree (16). API errors return with their kind and status, path errors are tool errors, and unknown categories come back as a notice (16). readOnlyHint and destructiveHint on every definition, and the grant filter tells the model never to run destructive tools autonomously (18). Official TypeScript and Python SDKs for the Neon API; a no-auth tool preview endpoint (14). 15 + 16 + 16 + 18 + 14 = 79.

PlanetScale MCP Server 74

26 tools since planetscale_get_branch_metrics arrived on 9 October (15). The insights-only endpoint drops the two SQL tools, and OAuth scopes limit which tool groups a client is granted (5). Insights, error, tag and log tools take limit with documented maxima and report truncated or has_next, and the logs tool pages. planetscale_execute_read_query returns every row with no cap (15). Error texts say what to do next, for example the safe-migrations message with two ways forward, the postgres_rls_active warning on empty results and the full-access message on a payment 403 (15). The 14 source tools set readOnlyHint and destructiveHint, with a test that fails on a missing annotation. Annotations on the 12 generated tools were not read. No idempotency for writes was found, though the payment tool tells the model not to start a second setup while one is pending (15). Sensible defaults (replica routing, 24-hour windows, curated metrics), but three identifiers are required on nearly every call. Official SDKs do not apply to a hosted MCP server and none were checked (9). Total 74.