---
title: "50 Best Jev Tools & Projects (2026 TypeSafe Directory)"
canonical_url: "https://www.calypso.so/es/ai-lists/best-jev-tools-and-projects"
last_updated: "2026-10-24T15:14:49.870Z"
locale: es
meta:
description: "The 50 best tools and projects built on TypeSafe's Jev: browser agents, Claude Code plugins, n8n and MCP integrations, semantic search, and open models."
keywords: "best Jev tools, Jev projects, TypeSafe Jev, TypeSafe AI, System One model, Jev MCP server, Jev n8n, Jev alternatives, open source Jev, Calypso"
"og:description": "The 50 best tools and projects built on TypeSafe's Jev: browser agents, Claude Code plugins, n8n and MCP integrations, semantic search, and open models."
"og:title": "50 Best Jev Tools & Projects (2026 TypeSafe Directory)"
"twitter:description": "The 50 best tools and projects built on TypeSafe's Jev: browser agents, Claude Code plugins, n8n and MCP integrations, semantic search, and open models."
"twitter:title": "50 Best Jev Tools & Projects (2026 TypeSafe Directory)"
---
Organizar mi WhatsApp
AI Lists / AI Tools
# **50 Mind-Blowing Jev Projects**
A customer asks for a refund. An agent needs its next action. A search returns ten plausible answers. Each creates a decision: where should this go, what should happen next, and is there enough evidence to proceed? Jev, TypeSafe’s structured decision model, helps applications make those choices. This September 2026 collection brings together 50 Jev tools, integrations, and implementation resources for conversation routing, inbox triage, browser automation, coding agents, and semantic search. Explore by use case, with official resources, community projects, and independent alternatives clearly identified.
## **Conversation Routing****7 tools ** Decide who should handle a message and what should happen next. These resources support the classification and workflow layers of messaging automation; WhatsApp delivery, conversation history, permissions, and team assignment remain part of your integration. **Copy link ** [**TypeSafe Intent Routing** An official recipe for classifying requests and choosing between application logic, a specialist model, and a human. Adapt this pattern to WhatsApp support routing, sales inquiries, returns, and order-status questions through your messaging integration.](https://docs.typesafe.ai/patterns/intent-routing "An official recipe for classifying requests and choosing between application logic, a specialist model, and a human. Adapt this pattern to WhatsApp support routing, sales inquiries, returns, and order-status questions through your messaging integration.")**0**[**TypeSafe Confidence-Gated Routing** An official pattern for deciding when to act, ask for clarification, or escalate. Useful for ambiguous messages and customer-service handoffs, with thresholds tested against your own conversations and the consequences of a wrong decision.](https://docs.typesafe.ai/patterns/confidence-routing "An official pattern for deciding when to act, ask for clarification, or escalate. Useful for ambiguous messages and customer-service handoffs, with thresholds tested against your own conversations and the consequences of a wrong decision.")**0**[**TypeSafe Community Nodes for n8n** Connect Jev decisions to visual automation workflows, including routing branches and a review output. A practical building block for message triage when combined with your own WhatsApp, CRM, or helpdesk integration.](https://github.com/itsmeyaw/n8n-nodes-typesafe "Connect Jev decisions to visual automation workflows, including routing branches and a review output. A practical building block for message triage when combined with your own WhatsApp, CRM, or helpdesk integration.")**0**[**Jevmail** A community Gmail triage app that groups messages into categories such as needs reply, updates, promotions, sales, and spam. Its read-only approach makes it a useful example of inbox classification without automatic sending.](https://github.com/fazlerocks/jevmail "A community Gmail triage app that groups messages into categories such as needs reply, updates, promotions, sales, and spam. Its read-only approach makes it a useful example of inbox classification without automatic sending.")**0**[**TypeSafe for LangChain** Bring structured TypeSafe decisions into LangChain applications. Relevant when an agent needs a classification or tool-risk assessment before continuing a workflow, invoking another model, or requesting human involvement.](https://docs.langchain.com/oss/python/integrations/providers/typesafe "Bring structured TypeSafe decisions into LangChain applications. Relevant when an agent needs a classification or tool-risk assessment before continuing a workflow, invoking another model, or requesting human involvement.")**0**[**Jev MCP** A community MCP server that exposes Jev judgments as tools for compatible assistants. Useful for adding explicit classification and scoring steps to an existing agent workflow.](https://github.com/jkudish/jev-mcp "A community MCP server that exposes Jev judgments as tools for compatible assistants. Useful for adding explicit classification and scoring steps to an existing agent workflow.")**0**[**System One Connector** A community MCP connector for TypeSafe System One models. Explore it when an MCP-enabled application needs structured evaluations alongside the model responsible for writing responses.](https://github.com/itsmostafa/system-one-connector "A community MCP connector for TypeSafe System One models. Explore it when an MCP-enabled application needs structured evaluations alongside the model responsible for writing responses.")**0**## **Model Routing****6 tools ** Put Jev inside an existing application stack, select downstream models, and inspect the resulting decisions. These integrations connect classification to deployment, evaluation, and debugging. **Copy link ** [**OpenRouter Jev Router** Uses Jev to select a model and reasoning effort for each request. A relevant option for applications balancing response quality, latency, and cost across different tasks within a conversation.](https://openrouter.ai/typesafe/jev-router "Uses Jev to select a model and reasoning effort for each request. A relevant option for applications balancing response quality, latency, and cost across different tasks within a conversation.")**0**[**TypeSafe for Vercel AI SDK** Use TypeSafe evaluations inside the AI SDK ecosystem. Useful for TypeScript applications that need a typed routing or scoring result before calling the model that generates the response.](https://ai-sdk.dev/providers/ai-sdk-providers/typesafe-ai "Use TypeSafe evaluations inside the AI SDK ecosystem. Useful for TypeScript applications that need a typed routing or scoring result before calling the model that generates the response.")**0**[**TypeSafe for TanStack AI** A TypeSafe adapter for TanStack AI’s structured decision APIs. Integrate choice, score, and boolean evaluations into applications that already use TanStack for their AI workflows.](https://tanstack.com/ai/latest/docs/adapters/typesafe "A TypeSafe adapter for TanStack AI’s structured decision APIs. Integrate choice, score, and boolean evaluations into applications that already use TanStack for their AI workflows.")**0**[**Jev in Cloudflare AI** Cloudflare’s model documentation for accessing Jev as a third-party provider. A useful integration reference for teams exploring structured decisions within a Cloudflare-based application stack.](https://developers.cloudflare.com/ai/models/typesafe/jev/ "Cloudflare’s model documentation for accessing Jev as a third-party provider. A useful integration reference for teams exploring structured decisions within a Cloudflare-based application stack.")**0**[**OpenInference TypeSafe Instrumentation** Instrument TypeSafe SDK calls with OpenTelemetry-compatible traces. Use it to inspect decision requests, connect them to wider agent activity, and investigate unexpected routing outcomes.](https://github.com/Arize-ai/openinference "Instrument TypeSafe SDK calls with OpenTelemetry-compatible traces. Use it to inspect decision requests, connect them to wider agent activity, and investigate unexpected routing outcomes.")**0**[**Langfuse Jev Evaluators** Evaluate observations with typed Jev questions inside Langfuse. Useful for repeatable verdicts and scoring criteria when you need an evaluation result rather than a written explanation.](https://langfuse.com/docs/evaluation/evaluation-methods/jev-as-a-judge "Evaluate observations with typed Jev questions inside Langfuse. Useful for repeatable verdicts and scoring criteria when you need an evaluation result rather than a written explanation.")**0**## **Browser Agents****5 tools ** Browser agents repeatedly choose an action, a target, and a next step. These projects explore Jev as that decision component, alongside separate tools for reading pages, generating text, and executing actions. **Copy link ** [**Browser Use Jev Ultrafast** A browser-agent implementation that uses Jev to choose actions and page elements from an indexed action space. A separate language model handles text entry, making the division of responsibilities easy to study.](https://github.com/browser-use/jev-ultrafast "A browser-agent implementation that uses Jev to choose actions and page elements from an indexed action space. A separate language model handles text entry, making the division of responsibilities easy to study.")**0**[**Jev Voice Browser** A community browser-control demo that combines speech transcription, page context, and Jev decisions. Explore voice-command routing while keeping transcription, intent interpretation, and browser execution as separate steps.](https://github.com/moritzkremb/jev-voice-browser "A community browser-control demo that combines speech transcription, page context, and Jev decisions. Explore voice-command routing while keeping transcription, intent interpretation, and browser execution as separate steps.")**0**[**Jev Browser Use for Codex** A community Codex skill that delegates browser navigation decisions to Jev. Codex retains responsibilities such as text generation, visual judgment, and final verification.](https://github.com/wy-coliney/jev-browser-use "A community Codex skill that delegates browser navigation decisions to Jev. Codex retains responsibilities such as text generation, visual judgment, and final verification.")**0**[**Foreman** A community agent-supervision project built around Jev. Explore how structured decisions can coordinate ongoing software work and determine what an agent should do next.](https://github.com/thruwire/foreman "A community agent-supervision project built around Jev. Explore how structured decisions can coordinate ongoing software work and determine what an agent should do next.")**0**[**Hermes Jev Skills** Community skills covering model routing, memory, context compaction, skill selection, and browser or computer workflows. Useful for studying several decision points within one agent system.](https://github.com/kerpopule/hermes-jev-skills "Community skills covering model routing, memory, context compaction, skill selection, and browser or computer workflows. Useful for studying several decision points within one agent system.")**0**## **Coding Agents****9 tools ** Find the relevant code, retain useful context, and choose how much model capability a task needs. These community tools apply Jev to the repeated decisions surrounding software development. **Copy link ** [**jevgrep** Search a codebase with natural-language questions and retrieve relevant files and verbatim excerpts. Useful when you know the behavior you need to investigate but not the filenames or symbols.](https://github.com/dzhng/jevgrep "Search a codebase with natural-language questions and retrieve relevant files and verbatim excerpts. Useful when you know the behavior you need to investigate but not the filenames or symbols.")**0**[**Fast Jev Compaction** A Claude Code plugin that uses Jev to decide which tool calls and results to retain, drop, or truncate. Retained content stays verbatim, helping preserve original evidence within a smaller context.](https://github.com/tamaratran/fast-jev-compaction "A Claude Code plugin that uses Jev to decide which tool calls and results to retain, drop, or truncate. Retained content stays verbatim, helping preserve original evidence within a smaller context.")**0**[**jev-review** A community tool for reviewing staged code changes with Jev and a local dashboard. Useful for experimenting with focused review checks before a commit.](https://github.com/devagrawal09/jev-review "A community tool for reviewing staged code changes with Jev and a local dashboard. Useful for experimenting with focused review checks before a commit.")**0**[**Jev Router for Claude Code** A community model router for Claude Code tasks. Explore assigning different requests to different models, then measure whether the routing improves your own balance of quality, speed, and cost.](https://github.com/gargpratyush/jev-router "A community model router for Claude Code tasks. Explore assigning different requests to different models, then measure whether the routing improves your own balance of quality, speed, and cost.")**0**[**Jev Codex Router** A community router that selects model, reasoning, and speed settings for Codex turns. Relevant when a coding session mixes straightforward edits with more demanding investigation.](https://github.com/0xNatoshi/jev-codex-router "A community router that selects model, reasoning, and speed settings for Codex turns. Relevant when a coding session mixes straightforward edits with more demanding investigation.")**0**[**jevmem** A community project-memory tool for coding assistants, including Claude Code, Cursor, and Codex. Explore retaining useful project context across sessions instead of repeatedly reconstructing it.](https://github.com/Avinash-jetwani/jevmem "A community project-memory tool for coding assistants, including Claude Code, Cursor, and Codex. Explore retaining useful project context across sessions instead of repeatedly reconstructing it.")**0**[**Jev Pruner** Uses Jev to trim lengthy Bash output before it reaches a coding model. A focused approach to reducing context noise from logs, command results, and verbose development tools.](https://github.com/tamaratran/jev-pruner "Uses Jev to trim lengthy Bash output before it reaches a coding model. A focused approach to reducing context noise from logs, command results, and verbose development tools.")**0**[**pi-jev** Adds Jev decisions to the Pi coding agent, including a tool-call gate and an explicit judgment tool. Useful for exploring decision checks around agent actions.](https://github.com/y0usaf/pi-jev "Adds Jev decisions to the Pi coding agent, including a tool-call gate and an explicit judgment tool. Useful for exploring decision checks around agent actions.")**0**[**Canny** Combines deterministic hooks, recorded evidence, and Jev advice around agent completion. Explore workflows that check what was actually done before accepting a coding task as finished.](https://github.com/qkal/Canny "Combines deterministic hooks, recorded evidence, and Jev advice around agent completion. Explore workflows that check what was actually done before accepting a coding task as finished.")**0**## **Search & RAG****8 tools ** Retrieving text is only the first step. These tools help classify records, rank candidates, and decide which evidence belongs in an answer, including knowledge-base responses for messaging and customer support. **Copy link ** [**pg-jev** A PostgreSQL extension for Jev-powered filtering, ranking, and classification. Explore questions such as which support records describe urgent issues or which team should review a ticket.](https://github.com/realZachi/pg-jev "A PostgreSQL extension for Jev-powered filtering, ranking, and classification. Explore questions such as which support records describe urgent issues or which team should review a ticket.")**0**[**Jev Search** Combines Search1API with Jev for query interpretation, source selection, and result ranking. Useful for studying the decisions between receiving a question and selecting material for an answer.](https://github.com/superagents-lab/jev-search "Combines Search1API with Jev for query interpretation, source selection, and result ranking. Useful for studying the decisions between receiving a question and selecting material for an answer.")**0**[**neo4jev** Explores Jev-guided traversal of a Neo4j graph. A useful reference for applications that must choose relevant relationships and neighboring records while investigating a question.](https://github.com/jexp/neo4jev "Explores Jev-guided traversal of a Neo4j graph. A useful reference for applications that must choose relevant relationships and neighboring records while investigating a question.")**0**[**jev-semgrep** A semantic command-line search tool that matches lines by meaning and supports logical combinations. Useful when exact keywords miss the wording used in logs, messages, or documents.](https://github.com/uehaj/jev-semgrep "A semantic command-line search tool that matches lines by meaning and supports logical combinations. Useful when exact keywords miss the wording used in logs, messages, or documents.")**0**[**jgrep** Filter text using a natural-language description of what you want to find. A compact example of bringing Jev classification into familiar command-line workflows.](https://github.com/keltokhy/jgrep "Filter text using a natural-language description of what you want to find. A compact example of bringing Jev classification into familiar command-line workflows.")**0**[**jev-seo** A Rust-based SEO and generative-search audit project with Jev-assisted analysis. Explore site checks and citation-oriented diagnostics as inputs to editorial improvements, without treating audit results as ranking guarantees.](https://github.com/AkashPriyadarshii/jev-seo "A Rust-based SEO and generative-search audit project with Jev-assisted analysis. Explore site checks and citation-oriented diagnostics as inputs to editorial improvements, without treating audit results as ranking guarantees.")**0**[**semdecide** Bring typed semantic decisions into Unix pipelines and CI workflows. Useful when a script needs to classify text before taking an explicit, testable branch.](https://github.com/sharziki/semdecide "Bring typed semantic decisions into Unix pipelines and CI workflows. Useful when a script needs to classify text before taking an explicit, testable branch.")**0**[**TypeSafe RAG Passage Classification** An official cookbook for checking retrieved passages for relevance, usable evidence, contradictions, and attempted instructions. Apply the pattern before a support assistant drafts an answer from a knowledge base.](https://docs.typesafe.ai/cookbooks/classifying_rag_passages "An official cookbook for checking retrieved passages for relevance, usable evidence, contradictions, and attempted instructions. Apply the pattern before a support assistant drafts an answer from a knowledge base.")**0**## **Official SDKs****7 tools ** Build the first working decision, inspect its behavior, and compare alternatives. Official clients and evaluation resources provide a foundation; community guidance adds implementation examples. **Copy link ** [**TypeSafe Console** The official account entry point for TypeSafe. Start here when setting up access to the service behind Jev, then use the SDK documentation to build your first evaluation.](https://console.typesafe.ai/ "The official account entry point for TypeSafe. Start here when setting up access to the service behind Jev, then use the SDK documentation to build your first evaluation.")**0**[**TypeSafe Python SDK** The official Python client for TypeSafe’s structured evaluations. A direct starting point for backend classification, ticket routing, batch processing, and experiments using your own examples.](https://github.com/typesafe-ai/typesafe-sdk-python "The official Python client for TypeSafe’s structured evaluations. A direct starting point for backend classification, ticket routing, batch processing, and experiments using your own examples.")**0**[**TypeSafe JavaScript and TypeScript SDK** The official JavaScript and TypeScript client. Use it to add typed decisions to server-side application workflows, including message classification and routing logic.](https://github.com/typesafe-ai/typesafe-sdk-js "The official JavaScript and TypeScript client. Use it to add typed decisions to server-side application workflows, including message classification and routing logic.")**0**[**Official TypeSafe Agent Skills** Agent skills published by TypeSafe for building with its System One API. Useful when a coding assistant needs implementation guidance for structured evaluation requests.](https://github.com/typesafe-ai/skills "Agent skills published by TypeSafe for building with its System One API. Useful when a coding assistant needs implementation guidance for structured evaluation requests.")**0**[**Building with Jev** A community skill for writing and improving Jev programs. Explore it for guidance on turning an application decision into explicit questions, criteria, and surrounding code.](https://github.com/dbreunig/building-with-jev-skill "A community skill for writing and improving Jev programs. Explore it for guidance on turning an application decision into explicit questions, criteria, and surrounding code.")**0**[**TypeSafe Evals** TypeSafe’s published evaluation dashboard. Use it to understand the vendor’s tested workflows and methodology, then benchmark the decisions that matter in your own application.](https://evals.typesafe.ai/ "TypeSafe’s published evaluation dashboard. Use it to understand the vendor’s tested workflows and methodology, then benchmark the decisions that matter in your own application.")**0**[**System One Adapter for Python** An official adapter that backs the TypeSafe evaluation interface with other model providers. Useful for comparing decision quality, cost, and latency while keeping the application interface consistent.](https://github.com/typesafe-ai/system-one-adapter-python "An official adapter that backs the TypeSafe evaluation interface with other model providers. Useful for comparing decision quality, cost, and latency while keeping the application interface consistent.")**0**## **Local Models****8 tools ** Explore the broader decision-model ecosystem through independent research and local runtimes. Similar names or compatible interfaces do not imply shared model weights, equivalent accuracy, or matching confidence behavior. **Copy link ** [**SemIf, formerly OpenJev** An independent project exploring structured decisions using an open model and probability readouts. Relevant for studying the decision-model approach; it is not an official release of Jev’s weights.](https://github.com/TheoLeeCJ/SemIf-OpenJev "An independent project exploring structured decisions using an open model and probability readouts. Relevant for studying the decision-model approach; it is not an official release of Jev’s weights.")**0**[**NanoJev** An independent research project exploring parallel decisions, dynamic candidate options, and a training pipeline. Useful for developers investigating how to build their own specialized decision models.](https://github.com/TianyuCodings/NanoJev "An independent research project exploring parallel decisions, dynamic candidate options, and a training pipeline. Useful for developers investigating how to build their own specialized decision models.")**0**[**AnyJev** A research implementation that places a Jev-style decision interface over language models without additional training. Useful for comparing structured decision approaches across supported model backends.](https://github.com/nokia-applied-research/AnyJev "A research implementation that places a Jev-style decision interface over language models without additional training. Useful for comparing structured decision approaches across supported model backends.")**0**[**Ollaya** A runtime for running decision models locally with a TypeSafe-compatible interface. Explore it when local deployment is a requirement, checking the capabilities and requirements of the selected model.](https://ollaya.dev/ "A runtime for running decision models locally with a TypeSafe-compatible interface. Explore it when local deployment is a requirement, checking the capabilities and requirements of the selected model.")**0**[**Von** An independent local decision-model project. Relevant for developers exploring self-hosted classification and measuring performance on their own hardware and task examples.](https://github.com/wfzyx/von "An independent local decision-model project. Relevant for developers exploring self-hosted classification and measuring performance on their own hardware and task examples.")**0**[**Simple Jev** An open-model approach to serving classifier-style decision endpoints. Useful for experimenting with the serving layer behind structured choices and comparing it with a hosted decision API.](https://github.com/featherless-ai/simple-jev "An open-model approach to serving classifier-style decision endpoints. Useful for experimenting with the serving layer behind structured choices and comparing it with a hosted decision API.")**0**[**OpenJev by razorback16** An independent DiffusionGemma-based server offering a Jev-compatible interface. A separate project from SemIf, suited to compatibility experiments rather than assumptions of identical model behavior.](https://github.com/razorback16/openjev "An independent DiffusionGemma-based server offering a Jev-compatible interface. A separate project from SemIf, suited to compatibility experiments rather than assumptions of identical model behavior.")**0**[**Valen** An experimental project exploring multimodal decision-model training. Relevant for research involving visual inputs; its capabilities should be evaluated separately from TypeSafe’s Jev service.](https://github.com/Liuziyu77/Valen "An experimental project exploring multimodal decision-model training. Relevant for research involving visual inputs; its capabilities should be evaluated separately from TypeSafe’s Jev service.")**0****The takeaway ****Editorial note. ** Start with one decision you can measure: which team owns a conversation, which document supports an answer, or which model should handle a request. For WhatsApp automation, your integration supplies the message and relevant context; Jev classifies or scores it; your application applies routing rules and executes the handoff. Keep reply generation separate, test decisions against real examples, and provide a review path for uncertain cases. The useful benchmark is whether the complete workflow becomes faster, more accurate, or less expensive.**Keep exploring **## Related AI lists. More curated directories from the same topic cluster so visitors can keep comparing tools and use cases. [FeaturedAI Tools** 100 tools****100 Mind-Blowing AI Tools**
A curated directory of standout AI tools across six categories — chatbots, marketing, writing, productivity, design, and video — to help teams find the right assistant for every workflow.Updated Jun 17, 2026** View list **](https://www.calypso.so/ai-lists/100-mind-blowing-ai-tools) [ai-search** 25 tools****25 Hottest RAG Tools for n8n AI Agents**
A curated directory of the 25 strongest RAG tools, n8n templates, vector-store nodes, and MCP building blocks for source-backed AI agents.Updated Jun 23, 2026** View list **](https://www.calypso.so/ai-lists/best-rag-tools-for-n8n-workflows-and-ai-agents)**From research to answers**## Turn trusted AI research into answers Calypso can cite. ### **Calypso WhatsApp** Organiza las conversaciones con clientes. Dale a cada pregunta un siguiente paso claro. [**~~Conoce WhatsApp ~~**](https://www.calypso.so/es/whatsapp) ### **Calypso Context** Crea respuestas con tu propio conocimiento, con citas que tus lectores pueden seguir. [**Consigue tu API key**](https://context.calypso.so/join)