THE COMMUNITY RESOURCE
DIRECTORY
Small decisions.
Small decisions.
Endless possibilities.
Find what people build with Jev. Explore tools, libraries, experiments and ideas from across the TypeSafe ecosystem.
11,066unique resources
29clear categories
220focused subcategories
Yourssave a personal reading list
YOUR NEXT BUILD STARTS HERE
Download catalog
Explore the collection
Browse groups
Frameworks & integrations11 of 335 matches
More projects & source code4 of 148 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| OpenClaw TypeSafe plugingithub.com | 390,115 | Official OpenClaw plugin that plugs hosted Jev, or a local Kev server, into OpenClaw's decision-model API for Choice, Score and Boolean judgments, plus an optional typesafe_evaluate tool. | Frameworks & integrationsMore projects & source code | |
| OpenViking rerankergithub.com | ≈38,400 | Rerank provider for a context database that scores every document with a Noul in one request. | Frameworks & integrationsMore projects & source code | |
| Composiogithub.com | 30,386 | TypeScript provider that picks and gates tool calls with a Choice and binds closed-set arguments before execution. | Frameworks & integrationsMore projects & source code | |
| ComposioHQ/composiogithub.com | 30,386 | Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action. | Frameworks & integrationsMore projects & source code |
AI frameworks7 of 75 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| langchain-typesafegithub.com | 146,708 | LangChain partner package with a TypeSafeClassifier runnable for Choice, Noul and Score questions, plus experimental auto-mode and model-router middleware. | Frameworks & integrationsAI frameworks | |
| AI SDK TypeSafe providergithub.com | ≈27,000 | Official AI SDK provider package @ai-sdk/typesafe-ai that runs Choice, Score and Boolean questions against Jev through the experimental evaluate API. | Frameworks & integrationsAI frameworks | |
| AI SDK TypeSafe provider — examplesgithub.com | ≈27,000 | Official AI SDK provider package @ai-sdk/typesafe-ai that runs Choice, Score and Boolean questions against Jev through the experimental evaluate API. | Frameworks & integrationsAI frameworks | |
| AI SDK TypeSafe provider — repogithub.com | ≈27,000 | Official AI SDK provider package @ai-sdk/typesafe-ai that runs Choice, Score and Boolean questions against Jev through the experimental evaluate API. | Frameworks & integrationsAI frameworks | |
| Pydantic AI TypeSafeModelgithub.com | 20,310 | Pydantic AI model class that runs decision agents on Jev: each field of the output_type becomes one question and the answers fill the output, so swapping the model name compares Jev with an LLM. | Frameworks & integrationsAI frameworks | |
| Pydantic AI TypeSafeModel — docsgithub.com | 20,310 | Pydantic AI model class that runs decision agents on Jev: each field of the output_type becomes one question and the answers fill the output, so swapping the model name compares Jev with an LLM. | Frameworks & integrationsAI frameworks | |
| LangChain.jsgithub.com | 18,244 | JavaScript version of the classifier and the routing and approval middleware. | Frameworks & integrationsAI frameworks |
Other resources & source leads2 of 156 matches
More projects & source code2 of 35 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| openclaw/openclawgithub.com | 390,115 | The AI that really does things. Any OS. Any Platform. The lobster way. 🦞 | Other resources & source leadsMore projects & source code | |
| Wei-Shaw/sub2apigithub.com | 42,112 | Sub2API 一站式开源中转服务，让 Claude、Openai 、Gemini、Grok订阅统一接入，支持拼车共享，更高效分摊成本，原生工具无缝使用。 | Other resources & source leadsMore projects & source code |
Agents & orchestration7 of 1,475 matches
More projects & source code5 of 959 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| AutoGPT TypeSafe blocksgithub.com | ≈187,500 | Seven no-code blocks, including a five-exit router, a yes, no, or unsure split, and a score filter. | Agents & orchestrationMore projects & source code | |
| BerriAI/litellmgithub.com | 59,200 | The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM] | Agents & orchestrationMore projects & source code | |
| composio — READMEgithub.com | 30,386 | Composio's TypeSafe adapter turns tools into Jev questions, then passes Jev's decision to Composio for execution. | Agents & orchestrationMore projects & source code | |
| pydantic-ai — typesafegithub.com | 20,310 | The Pydantic AI repository contains a TypeSafe provider and TypeSafeModel backed by the TypeSafe SDK for Jev typed decisions; this is one specific integration, not a claim that the whole framework runs on Jev. | Agents & orchestrationMore projects & source code | |
| pydantic-ai — typesafe.pygithub.com | 20,310 | The Pydantic AI repository contains a TypeSafe provider and TypeSafeModel backed by the TypeSafe SDK for Jev typed decisions; this is one specific integration, not a claim that the whole framework runs on Jev. | Agents & orchestrationMore projects & source code |
Model & task routing2 of 139 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| LiteLLM Jev complexity routergithub.com | 59,200 | LiteLLM's complexity-based router can use Jev to classify each request into configured tiers that pick the backend model; a separate guardrail blanks tool results Jev judges no longer needed. | Agents & orchestrationModel & task routing | |
| LiteLLM Jev complexity router — guardrailgithub.com | 59,200 | LiteLLM's complexity-based router can use Jev to classify each request into configured tiers that pick the backend model; a separate guardrail blanks tool results Jev judges no longer needed. | Agents & orchestrationModel & task routing |
SDKs & client libraries3 of 1,129 matches
More projects & source code3 of 689 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| langchain-typesafe — repogithub.com | 146,708 | LangChain partner package with a TypeSafeClassifier runnable for Choice, Noul and Score questions, plus experimental auto-mode and model-router middleware. | SDKs & client librariesMore projects & source code | |
| Pydantic AI TypeSafeModel — repogithub.com | 20,310 | Pydantic AI model class that runs decision agents on Jev: each field of the output_type becomes one question and the answers fill the output, so swapping the model name compares Jev with an LLM. | SDKs & client librariesMore projects & source code | |
| json-render Jev composer — repogithub.com | ≈18,400 | Experimental json-render playground mode where Jev composes and edits UI trees from a shadcn component catalog through batches of Choice questions on which elements to include, where they go and in what order. | SDKs & client librariesMore projects & source code |
Open models & compatible servers2 of 569 matches
More projects & source code2 of 217 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| vLLM DiffusionGemma Jev modegithub.com | 92,393 | Open vLLM pull request adding a structured generation mode that turns DiffusionGemma into a Jev-like model, with a prototype server for the /v1/systemone endpoint. | Open models & compatible serversMore projects & source code | |
| cua-s1-forms — repogithub.com | ≈26,700 | Tiny Jev-like option scorer (706,048 parameters, 2.8 MB) that rates fill, check, click or skip for every form field in one parallel pass, built as the decision layer for cua-driver. | Open models & compatible serversMore projects & source code |
Data, classification & extraction4 of 218 matches
Classification & labeling2 of 48 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| World Monitor Jev headline classifiergithub.com | 87,060 | Headline classifier in a real-time geopolitical news dashboard that asks jev-1.13.0 for a five-level severity and one of 14 topic categories per headline, validating answers and falling back on failure. | Data, classification & extractionClassification & labeling | |
| World Monitor Jev headline classifier — repogithub.com | 87,060 | Headline classifier in a real-time geopolitical news dashboard that asks jev-1.13.0 for a five-level severity and one of 14 topic categories per headline, validating answers and falling back on failure. | Data, classification & extractionClassification & labeling |
More projects & source code2 of 102 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| Airflow LLM branchinggithub.com | ≈46,900 | Picks the next task with a Jev Choice and hands low-confidence cases to a person. | Data, classification & extractionMore projects & source code | |
| Airflow LLM branching — repogithub.com | ≈46,900 | Picks the next task with a Jev Choice and hands low-confidence cases to a person. | Data, classification & extractionMore projects & source code |
Benchmarks & evaluation1 of 917 matches
Evaluation harnesses1 of 39 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| World Monitor Jev Headline Evaluationgithub.com | 87,060 | Experimental harness that scores news headlines against geopolitical threat levels and event categories. | Benchmarks & evaluationEvaluation harnesses |
Finance & trading2 of 152 matches
Trading agents2 of 48 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| AI Hedge Fund Jev adaptergithub.com | 63,580 | Native Jev adapter in the educational AI Hedge Fund project, letting its investor agents run a mandate with jev-1.13.0 and turning typed answers into the same JSON contract the chat models return. | Finance & tradingTrading agents | |
| AI Hedge Fund Jev adapter — repogithub.com | 63,580 | Native Jev adapter in the educational AI Hedge Fund project, letting its investor agents run a mandate with jev-1.13.0 and turning typed answers into the same JSON contract the chat models return. | Finance & tradingTrading agents |
Customer support & sales1 of 53 matches
More projects & source code1 of 4 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| Twenty Classify workflow actiongithub.com | ≈57,200 | Classify step in the open-source Twenty CRM's workflows that asks Jev choice, score or boolean questions about a record, so later steps can branch on the answers and probabilities. | Customer support & salesMore projects & source code |
Safety, moderation & guardrails1 of 244 matches
Content moderation1 of 64 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| Sub2API content auditgithub.com | 42,112 | Content-moderation engine in the Sub2API relay that screens traffic with Jev across 13 categories such as harassment, hate, self-harm and violence, managed from an admin risk-control page. | Safety, moderation & guardrailsContent moderation |
Coding & developer tools5 of 1,328 matches
More projects & source code4 of 833 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| Oh My Pi Jev judgmentsgithub.com | 32,042 | Native Jev judgment backend in the Oh My Pi coding agent: a judge() call in its eval kernel and a jevify keyword that bulk-classifies diffs, logs or search hits so the agent reads only flagged items. | Coding & developer toolsMore projects & source code | |
| Oh My Pi Jev judgments — repogithub.com | 32,042 | Native Jev judgment backend in the Oh My Pi coding agent: a judge() call in its eval kernel and a jevify keyword that bulk-classifies diffs, logs or search hits so the agent reads only flagged items. | Coding & developer toolsMore projects & source code | |
| Jev skill suggestion for Claude Codegithub.com | 30,825 | Hides the skill roster and lets Jev attach at most one skill to each prompt. | Coding & developer toolsMore projects & source code | |
| Dub malicious link checkgithub.com | ≈24,800 | Screens every new short link on the Dub link platform with Jev after a domain blacklist check, blocking phishing, malware, cloaking redirectors, gambling and adult destinations. | Coding & developer toolsMore projects & source code |
Coding model routers1 of 106 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| Jev model router for Claude Codegithub.com | 30,825 | Hook that asks for a tier Choice, an effort Score, and a risk Noul in one request, then adjusts model and reasoning effort. | Coding & developer toolsCoding model routers |
MCP servers & agent skills1 of 93 matches
More projects & source code1 of 36 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| Jev skill suggestion for Claude Code — repogithub.com | 30,825 | Hides the skill roster and lets Jev attach at most one skill to each prompt. | MCP servers & agent skillsMore projects & source code |
Browser, mobile & computer use5 of 416 matches
Web browser agents3 of 119 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| Cua jev-usegithub.com | ≈26,700 | Public-preview recipe in Cua Driver where Jev chooses the next browser action from bounded candidate IDs while the driver observes the page, performs the action and verifies the result, in Python and TypeScript. | Browser, mobile & computer useWeb browser agents | |
| Cua jev-use — repogithub.com | ≈26,700 | Public-preview recipe in Cua Driver where Jev chooses the next browser action from bounded candidate IDs while the driver observes the page, performs the action and verifies the result, in Python and TypeScript. | Browser, mobile & computer useWeb browser agents | |
| jev-ultrafastgithub.com | 21,637 | Browser agent that picks each step's operation and target from an element table in one request, with a speculative target per operation and a small LLM only for typed text. | Browser, mobile & computer useWeb browser agents |
More projects & source code2 of 197 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| endpointgithub.com | 21,637 | browser-use/jev-ultrafast — i. am. speed. · endpoint · Python | Browser, mobile & computer useMore projects & source code | |
| jev-ultrafast — READMEgithub.com | 21,637 | The browser-use agent asks TypeSafe Jev to choose an operation and element; a separate small LLM writes text only for TYPE_TEXT. | Browser, mobile & computer useMore projects & source code |
Search, retrieval & RAG2 of 193 matches
Ranking & reranking2 of 50 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| Hindsight Jev rerankergithub.com | ≈24,900 | Reranker provider in the Hindsight agent-memory system that asks Jev one question with every recall candidate as an option, so the answer is the ranking, in a single request. | Search, retrieval & RAGRanking & reranking | |
| Hindsight Jev reranker — docsgithub.com | ≈24,900 | Reranker provider in the Hindsight agent-memory system that asks Jev one question with every recall candidate as an option, so the answer is the ranking, in a single request. | Search, retrieval & RAGRanking & reranking |
Writing, media & creative tools1 of 173 matches
More projects & source code1 of 41 matches
| Resource | Stars | Description | Category | Save |
|---|---|---|---|---|
| json-render Jev composergithub.com | ≈18,400 | Experimental json-render playground mode where Jev composes and edits UI trees from a shadcn component catalog through batches of Choice questions on which elements to include, where they go and in what order. | Writing, media & creative toolsMore projects & source code |
No matches this time
Try a broader search or reset your filters to explore the full collection.
Showing 48 of 11,066 resources