Discover the Best AI Tools

A prototype MCP gateway exposes paid text tools, model inference, Kaspa data and covenant or proof services; prepaid sessions are custodial, while optional mainnet covenant channels are unaudited and need a separate client for funding.

Switchboard is an OpenAI-compatible Python gateway using Jev to screen requests for injection, harm and PII and route to small or large models, with optional output checks. If Jev is unavailable it defaults to allowing the large model with a flag, unless configured fail-closed. Credentials are required; thresholds and security effectiveness need independent validation.

CyberWareX MCP provides agent access to web extraction, screenshots, PDF, blockchain data, speech and other hosted APIs. Calls are paid through x402 with USDC on Base. Servers return payment requirements rather than holding wallet keys or paying themselves; an authorized client handles settlement.

Ailerix is a model-routing prototype with an OpenAI-shaped completion API: Jev classifies task families and code selects from a cost/quality catalog. Development uses a fixed benchmark snapshot and local or mocked responses until service keys are configured. It is not evidence of a fully operational hosted routing service.

CortexCloud MCP exposes hosted AI generation, embeddings, search, market data and blockchain RPC through an MCP gateway. Calls are paid individually in USDC on Base using x402 and require an authorized funded wallet. No API key does not mean free access; upstream availability and fees apply.

An experimental Hono router that uses Jev to choose a matching route from request meaning. Semantic routing sends request data to an external service and adds latency and cost; it is not an authentication or authorization boundary.

klanex MCP connects agents to a hosted asynchronous API-execution service with schema checks, retries, status queries and approval gates. The local package is a stdio proxy, not the execution backend; a klanex key and service access are required.

HookSense MCP creates callback endpoints, waits for asynchronous results and exposes signature-verification tools to agents. It requires a HookSense account token and reachable external service; callback verification does not establish that the originating job completed correctly.

Yueli DEX combines local decision rules, scenario templates and provider routing while preserving Jev’s Choice contract. It supports Choice only and returns action intentions without executing them; packages currently build from source, with npm publication pending.

Ollama-Omega exposes Ollama health, model listing, chat, generation and model management through MCP. It requires a running Ollama daemon; models may be local or cloud-backed, and the bridge does not provide memory, authentication or policy enforcement.

Jev Switch routes typed Jev requests through configurable provider graphs using a Rust daemon, with web and desktop controls. Upstream services perform inference; local mode is not necessarily offline, and the public interface is Jev rather than a general Chat Completions API.

x402-mcp discovers paid APIs from CDP Bazaar and exposes them to agents while handling x402 USDC payments. It requires a funded wallet and signing credentials; the configurable per-call price cap is not a total spending budget.

typesafe-ai-rails wraps the community Ruby TypeSafe SDK with Rails configuration, persisted usage estimates and opt-in confidence policies for Choice and Score. It is unofficial and needs API credentials; Noul decisions use their probability directly rather than those confidence policies.

This NewAPI task plugin routes native TypeSafe System One requests to TypeSafe or Vercel AI Gateway, preserving typed answers and mapping usage for billing. It requires an upstream key and configuration; it does not implement chat-completions or Responses compatibility.

Laravel TypeSafe Jev adds Laravel configuration, scoped dependency injection, a facade and a recording test fake around a community PHP TypeSafe SDK. It is unofficial, requires API credentials for real calls, and handles typed decisions rather than chat generation.

Decis is a self-hostable inference server exposing a TypeSafe System One-compatible API for open decision models, with interchangeable engines and a playground. It is pre-1.0 software and does not run the proprietary Jev model.

Native C++ DuckDB extension batching and streaming Jev classification, predicates and rubric scores from SQL. Requires a TypeSafe key and matching extension build; distributed targets are macOS/Linux on x86-64 and ARM64.

Preview native MySQL plugin for natural-language row filtering and text comparisons through Jev. Requires a compatible Linux MySQL server, plugin-install privileges and a TypeSafe key; evaluated values and prompts leave the database for the API.

Neon Function showcase that judges request bodies with Jev and optionally forwards passing requests to a chosen HTTPS target. Self-deployment needs TypeSafe credentials; it is a demonstration gateway, not a hosted chat model.

MCP server indexing OpenAPI specifications with local embeddings and dependency graphs to discover and execute relevant endpoints. Local search needs no embedding API key; authenticated target APIs still require their credentials.

Pre-alpha PostgreSQL extension exposing TypeSafe Choice, Noul and Score through SQL, including batched calls. Requires PostgreSQL 15–18 on Linux or macOS, libcurl and a TypeSafe API key; it is an independent integration.

Self-hostable API server for text, vision and data tasks with local or hosted model backends. Tool-using agents need an OpenAI or OpenRouter tool-calling model; bundled business scenarios use demo data.

PostgreSQL extension using TypeSafe Jev to filter, rank and classify rows with natural-language conditions. Requires PostgreSQL 14–17, plpython3u, superuser access and an API key; restricted managed databases cannot run it.

MCP server for authenticated REST APIs with automatic login, configurable two-factor flows, Swagger retrieval and endpoint search. Requires API credentials and configuration; agents can make write requests, not just read data.