Jev directory
CommunityJev Integrations & Adapters
Pm Jev
A pm-cli extension and library that adds typed Jev decisions to project-management items. Commands cover triage, link, dedupe, ask, explain, gate, models, and doctor, with ranking, thresholds, and retries handled in code.
- Reading time
- 2 min
- Published by
- Community
- Added
- Nov 2026
Highlights
- Registers triage, link, dedupe, ask, explain, gate, models, and doctor commands with the pm host.
- Requires Node 22.18 or newer, pm-cli 2026.10.10 or newer, and Ollama with the /v1/systemone API.
- Defaults to the local tev1:4b model; there is no standalone bin, commands run through npx pm or bunx pm.
- Apply paths use --min-confidence thresholds, and dedupe proposals never close items.
- Default text output rounds probabilities to three decimals; --json keeps full precision and every option.
Questions about this entry
Short answers drawn from the linked project or page.
Install the package and a compatible pm host, then activate the extension in an initialized pm project: npm install pm-jev @unbrained/pm-cli, followed by npx --no-install pm package install ./node_modules/pm-jev. The digest notes the first npm release is pending the maintainer's interactive bootstrap.
No. The package exports a library and a pm extension and declares no standalone bin. Commands run through the installed host with npx --no-install pm jev ... or bunx --bun --no-install pm jev ...; installing the npm or Bun dependency alone does not activate it.
Node 22.18 or newer, pm-cli 2026.10.10 or newer, and Ollama with a decision-capable model exposing the /v1/systemone API. Bun is also supported as a host runtime.
The default provider is local Ollama with tev1:4b, which can be pulled with ollama pull tev1:4b. The pm jev models command lists decision models and the discovery source.
It diagnoses configuration, privacy opt-in, reachability, and model availability without sending item state.
Gate output uses holds for probability >= threshold, where equality counts as true, expected for --expect true|false (default true), and ok for holds === expected. The comparison uses full precision before display rounding, and exit status is 0 when ok, otherwise 1.
Triage, link, and dedupe runs without --apply make no item or history writes. With --apply they add qualifying changes and a receipt comment; dedupe proposals never close items, and hierarchy links require --allow-parent.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Agent Handoff Gate: agents verify what they hand off
X: Jev as an LLM guardrail inside Agentgateway
X: Vercel's fx safety reviewer, measured at 18× faster
Reddit: Early experiments wiring Jev into Pi as a safety layer and model router.
More Jev resources like this
Oxlint JS plugin that turns plain-English yes/no questions into lint rules answered by TypeSafe Jev. Rules target a function, call, JSX element, or whole file, and report when Jev's yes-probability clears a per-rule cutoff.
Jev Integrations & Adapters#javascript#typescript#guardrails
Community
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
Community n8n node that sends text or JSON items to Jev decision models and routes them by the answer. Supports Choice, Score, and Noul questions, backed by either a local Ollaya daemon or the hosted TypeSafe AI API.
Jev Integrations & Adapters#routing#local#typescript
Community
Effect provider package that serves the DecisionModel capability with Jev, the System One model answering typed questions with probabilities instead of text. Defines classify, rate, and probability decisions answered in one call, with typed AiError mapping.
Jev Integrations & Adapters#typescript#sdk#classification
Community
npm package that plugs TypeSafe Jev into Scruple as a decision provider. Scruple calls it for final rule decisions and for bounded collection classifications where a rule needs semantic help choosing candidates.
Jev Integrations & Adapters#typescript#routing#guardrails
Community
Optional TypeSafe AI Jev decision provider for the Zhin chat framework. Evaluates named Choice, Score and Noul questions through a plugin-managed client, returning labels, distributions and confidence without generating chat replies or granting tool permissions.
Jev Integrations & Adapters#typescript#sdk#routing
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.