Jev directory
CommunityJev Integrations & Adapters
@loopstack/typesafe
NestJS module that adds TypeSafe AI decisions to Loopstack workflows. It asks yes/no, choice, and score questions about workflow or agent state and returns probabilities, confidence values, and a legend per answer.
- Reading time
- 2 min
- Published by
- Community
- Added
- Nov 2026
Highlights
- Provides noul (yes/no), choice (pick-one-label), and score (rubric) question helpers.
- Answers carry probabilities, confidence, and a legend where applicable.
- Exposes a typesafe_system_one tool that agents can list in an LLM call's tools.
- Model defaults to TYPESAFE_DEFAULT_MODEL, then jev-latest; key read from TYPESAFE_API_KEY.
- Re-exports SDK error classes such as RateLimitError and AuthenticationError.
Questions about this entry
Short answers drawn from the linked project or page.
Install it with npm install @loopstack/typesafe, then import TypeSafeModule from the package and add it to the imports array of your NestJS app module.
Three: noul for yes/no questions, returning the probability of yes from 0 to 1; choice for pick-one-label questions, returning the choice, confidence, and probabilities; and score for rubric questions, returning a score, confidence, legend, and probabilities.
The API key is read from the TYPESAFE_API_KEY environment variable by default, and the envApiKey config option names a different variable. The model defaults to TYPESAFE_DEFAULT_MODEL, then to jev-latest.
Yes. Inject TypeSafeSystemOneTool and list typesafe_system_one in an LLM call's tools array so the agent can ask typed questions about state during a run.
A result of { model, answers, usage } with one answer per question name, plus metadata of { provider: 'typesafe', model, usage: { inputTokens, outputTokens } }.
The tool validates at runtime, so answers are typed per question kind, for example choice as string. TypeSafeClientService calls systemOne() directly and infers answer types from your own labels, such as 'billing' | 'other'.
It is built for the Loopstack automation framework and registers as a NestJS module, so it is not a fit for projects outside that stack or for standalone command-line use.
More Jev resources like this
LangChain.js integration for TypeSafe System One models. It wraps the jev classifier as a Runnable that returns typed choice, score, and noul answers with calibrated probabilities, plus agent middleware for model routing and tool gating.
Jev Integrations & Adapters#typescript#sdk#routing
Community
TypeScript package that implements the IDecisionModel interface for @mcp-abap-adt/llm-agent on top of @typesafe-ai/sdk, letting an agent ask Jev typed yes/no, choice, and score questions about a state and get numbers back.
Jev Integrations & Adapters#typescript#sdk#agents
Community
npm package providing the TypeSafe Jev fast path for the MLola Browser Runtime: one request per decision cycle carrying the operation head plus speculative target heads, with responses validated before anything executes. Also ships planner and text adapters for OpenAI-compatible endpoints.
Jev Integrations & Adapters#typescript#browser#agents
Community
npm package that decides what happens to each new fact in an agent memory store: add, update, supersede, or skip. It calls TypeSafe Jev for typed answers with probabilities and ships as part of Oort, open-source memory for AI agents. Apache-2.0.
Jev Integrations & Adapters#typescript#agents#sdk
Community
First-party TypeSafe AI decision provider for Standard Agents. It implements DecisionProviderInterface with kind 'decision' and decide(), sending explicit context and multiple independent questions to Jev in one request.
Jev Integrations & Adapters#typescript#sdk#agents
Community
n8n community node that adds a TypeSafe resource to workflows, exposing Get Many Models and Ask Questions operations against the TypeSafe API. Generated from OpenAPI 0.2.0 and authenticated with a bearer token credential.
Jev Integrations & Adapters#typescript#classification#sdk
Community
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.