Human escalation
Use calibrated confidence to decide what a person must see.
73 rows are filed under this pattern, 70 of them with code. Rows are listed as the catalogue orders them: official first, then rows with code, then by stars.
Filter and search them in the catalogue →
Rows marked “not Jev itself” or “shadow mode” stay listed, but they are not examples of deciding with Jev.
-
Cookbook: Classification using confidence
Classifies annual reports into 75 industry groups, then reads the answer's own confidence to decide whether to report that group or the broader division above it.
-
Cookbook: Double-checking citations
Catches wrong or invented citations against the source document with one Choice, using its confidence to flag borderline cases for review.
-
Cookbook: Knowledge graph entity alignment
Decides which of 450 candidate pairs from two product catalogues describe the same thing, with one Score whose three levels are the three available actions.
-
Cookbook: Self-consistency with choices
Adds an explicit "uncertain" outcome to moderation decisions and measures label agreement against the share of actions taken automatically.
-
Cookbook: Self-consistency with nouls
Routes uncertain probabilities to human review while keeping the underlying noul values visible rather than collapsing them to a label.
-
Pattern: Confidence-gated routing
Treat confidence as a second axis: the answer tells you what, the confidence tells you whether to act on it.
-
Confidence
How Choice and Score confidence is derived from probabilities, how Noul differs, and how to tune thresholds for the question and application.
-
Airflow LLMBranchOperator with Jev
Turns downstream task ids into a choice option set, with a minimum-confidence gate that routes uncertain runs to a human.
-
Composio TypeSafe provider
Compiles a tool catalogue into questions and reconstructs tool calls from the answers, with typed errors for abstention and confirmation-required cases.
-
Inbox Zero: seven email decisions
Seven distinct email decisions, each with its own separately chosen threshold, falling back to the normal LLM on any error.
-
jev-review
Pre-screens code review with Jev to surface high-risk changes for a more expensive model or a person, with a local dashboard.
-
jev-align
Builds calibrated decision functions from human feedback.
-
jev-forge
Not Jev: a toolkit that synthesizes decision data, trains and calibrates small candidate scorers (JevForge-0.8B) and serves them behind a Jev-compatible API, first for picking the next click or form control on a web page. Its benchmark sends the same held-out choice, noul and score questions to Jev.written by a model
-
neurolink
The pipe layer of an AI nervous system: one interface connecting provider neurons to an application, across three inference types — generate, stream, and decide. Decide returns typed, calibrated judgments (boolean/choice/score) via TypeSafe Jev or the open-weights Laya, not text.
-
Jev-Code-Reviewer
Local reviewer for coding agents' pull requests: each change is classified P0, P1 or P2 by Jev choices against a written priority policy, and only P0 changes are shown by default. Posts nothing to GitHub.written by a model
-
Jev-Moderation-Bot
A Discord moderation bot: a Choice tiers each message while a Noul carries ban urgency, and an admin pardon is fed back as a safe precedent in later requests.
-
jev-calibrate
Calibrate Jev questions against your own labels: tune criteria on labelled examples, confirm on a held-out set, get a verdict per question. Unofficial.upstream description
-
jev-agent-skill-router
Typed, confidence-aware agent skill routing with TypeSafe Jev.upstream description
-
toolgate
Open auto mode for AI agents — a calibrated tool-call firewall powered by TypeSafe Jev. Ships as a Claude Code hookearlier upstream description
-
jev-benchmarks
Probability-aware evaluation for typed decision models: calibration, selective risk, latency, and reproducible benchmarks.upstream description
-
jev-usecases
Production TypeSafe Jev (System One) use-case harnesses with confidence-gated decision logicupstream description
-
muse-jev-playbook
Jev decision layer for Muse: a fast, cheap TypeSafe AI gate before expensive agent work — confidence policy, recipes, reference router, honest measurement.upstream description
-
jeval
Measures what your Jev classifier's confidence is really worth, and sets the human hand-off line from what a mistake costs.earlier upstream description
-
jevwire
Jev decision layer for agents: MCP server, embeddable DecisionModel library, and an escalate-only Claude Code plugin (TypeSafe AI's Jev)upstream description
-
jevalyn
The decision layer for your Rails app. A Rails-native wrapper around TypeSafe's Jev System One API: typed, calibrated decisions in your control flow.upstream description
-
discern
Craft Type-Safe Uncertainty-aware semantic pattern matching, control flow, and smart procedures for Effect DecisionModel and Jevupstream description
-
jev-dspy-lab
Reproducible calibration and selective-risk benchmarks for Jev/TypeSafe decisions in DSPy workflowsupstream description
-
jev-harness
Decision harness for TypeSafe Jev — confidence gates, shadow mode, recipes, and evals. Claude CLI 48.9s → Jev 1.3s on the same row-filter job.upstream description
-
jev-engineering
Claude Code PreToolUse gate: regex hard rules and a read-only allowlist first, then one Noul and one Choice decide allow, ask or deny, so clearly safe commands skip the permission prompt. Codex, Cursor and OpenCode adapters, plus a review-router GitHub Action.
-
jevflow
Probabilistic AI decisions as composable backend primitives — typed judgments (noul/score/choice), deterministic thresholds, and explainable workflows. Powered by TypeSafe's Jev, provider-agnostic.upstream description
-
jevmory
Local memory for coding agents: hooks store what sessions actually said as verbatim quotes, and jevmory audit asks Jev, for each line of a MEMORY.md, whether stored evidence supports or contradicts it, marking it keep, stale, wrong or unsupported. Python standard library, SQLite; grading is opt-in.written by a model
-
poorjev
Open-source, local Jev alternative: a System One decision layer with provably calibrated confidence (ECE 0.170→0.071). Typed decisions, runs offline, no API key, no waitlist.upstream description
-
jev-phishing-bench
Jev (TypeSafe) vs Claude Haiku 4.5 on 2 000 phishing emails: accuracy, calibration, latency, cost. Reproducible benchmark.earlier upstream description
-
jevcal
Calibrate, threshold and drift-check a decision model against an LLM teacher instead of guessing a cutoff.
-
typed-decisions
Jev-shaped typed-decision model (state + Choice/Score/Noul questions -> calibrated probabilities, one pass) on ModernBERT / DeBERTa / LLaDA-MoE, with measured latency, accuracy, calibration and training costupstream description
-
jev-browse
Helpers for browser-harness that let Claude Code or Codex hand off a whole website sub-task in one fast_run call, with each step decided by Jev and a typed reason returned when it must stop. A port of browser-use/jev-ultrafast; the author reports 1.6–3.0x faster and 2.1–5.3x cheaper runs.written by a model
-
jev-block-android-ad
JevNoiseGate filters unwanted notifications and SMS on Android. Rather than matching keywords, an LLM decides what's noise — and only what it explicitly flags is blocked. Verification codes are matched on-device and never uploaded; anything uncertain passes through.upstream description
-
luce
Luce: a recipe for calibrated decision models — a sentence about your task in, a small model that answers typed questions with honest probabilities out (init → synth → train → eval → serve)upstream description
-
typesafe-local
Inspired by TypeSafe Ai, Ask a local LLM typed questions, get calibrated probabilities instead of text. Structured output without generation or parsing. MLX / Apple Silicon.upstream description
-
daf-jev
daf-jev: composable Python toolkit for TypeSafe's Jev (System One) decision API — question builders, confidence gates, evaluator, calibration, CLI, MCP server, agent skillupstream description
-
opencode-jev-orchestrator
Keeps OpenCode on a cheap sticky model for warm cache; Jev escalates hard turns to stronger subagents.upstream description
-
system-one-gemma
Open-source Jev-style System One decision model. Gemma 3 270M with a scoring head — fast, calibrated decisions in a single forward pass. No text generation. Inspired by TypeSafe.ai's Jev.upstream description
-
jev-ood-calibration
Independent calibration test of TypeSafe's Jev on a task it cannot have seen: 900 rule-generated support tickets (choice / score / boolean) plus 3 public benchmarks via Vercel AI Gateway. Raw responses, ECE with noise floor, temperature refit, per-type sign of miscalibration. Reproducible for ~
-
jev-ui
React components that resolve which component to render, how to order a list, and whether to show an affordance — from calibrated judgments returned by TypeSafe's Jev.upstream description
-
qwen-rlcd
Jev-style calibrated decision model (Choice/Score/Noul) on Qwen3.5-0.8Bupstream description
-
tink-route
Dynamic, confidence-aware Agent Skill routing with TypeSafe Jev and Tinkupstream description
-
jev-logtriage
Jev decides whether a batch of logs is worth acting on. Typed questions, confidence gates, nothing executed.upstream description
-
jev-starter
Typed, policy-driven decision workflows on top of TypeSafe AI Jev: confidence routing, fallbacks, evaluation, and RAG patterns for TypeScript apps.upstream description
-
solvi
Python runtime with a System One API adapter: Jev, an LLM or a local checkpoint proposes typed answers; hard checks, rules and calibrated thresholds decide, abstain or escalate to a person, and every decision is a hash-chained replayable trace.
-
tenbin
MCP server and agent skill for the TypeSafe AI System One API (Jev): decompose a judgment into Choice / Score / Noul questions, lint them, measure on labelled data, and put calibrated thresholds in codeupstream description
-
jev-flash-router
open-sourced jev-flash-router: an MCP server for TypeSafe's new Jev model. AI coding agents waste hundreds of reasoning tokens just deciding which file to edit, which route to pick, or whether a diff breaks tests. Jev evaluates state and outputs calibrated probabilities. Works with Cursor, Wind
-
jev-mcp-server
MCP server for Jev (TypeSafe System One): the three official question types — choice, score, noul — plus batch classify. Calibrated probabilities, ~0.5s, <$0.001/call.upstream description
-
jevbus
A streaming event bus whose routing, subscription and consumption are decided by a probabilistic judge. The reference judge is TypeSafe AI's Jev (System One) model: send it a payload and a set of typed questions, get back calibrated probabilities instead of prose.upstream description
-
pi-typesafe-jev
A pi extension that exposes TypeSafe (Jev, System One) judgments as five pi tools, so a model can make narrow semantic judgments while your code and your users keep control of thresholds, weights, and actions.upstream description
-
grok-jev-guard
A typed preflight and approval layer for Grok Bot: local policy owns the hard boundaries, Jev judges the ambiguous cases, and Grok Bot executes within the envelope it gets back.
-
opencode-jev-guard
OpenCode 2 plugin that sends every shell command (local or via FarHand) to TypeSafe's Jev and asks you first when it leaves files outside the project, installs software globally, changes global settings, is harmful or exposes private dataupstream description
-
qualm
Typed decisions from a System One model, where uncertainty is something you have to handle.upstream description
-
toolgate
Agent tool/MCP call gate — allow / ask_human / deny via TypeSafe Jevupstream description
-
watfile
Text/PDF - File categorization and sorting with Typesafe AI Jev or local calibrated decision modelupstream description
-
XavierJev
A local decision layer in Jev's shape: yes/no, choice and rubric questions read off one token's logprobs from a local model, with a Claude Code permission gate measured on held-out command sets.
-
jev-calibration-audit
Independent API-only calibration audit of TypeSafe AI's Jev decision modelupstream description
-
jev-eval
Benchmark TypeSafe Jev against any OpenRouter model on your own labelled classification data: accuracy, calibration, latency, cost
-
jev-guard
Vets an LLM agent's tool calls through TypeSafe AI's Jev before they run — allow, block, or hold, failing safe on uncertainty.upstream description
-
jev-review
Autonomous Jev pull-request review with typed decisions, calibrated approval gates, and trusted-owner escalationupstream description
-
jev-the-janitor
A janitor for markdown vaults powered by TypeSafe Jev: Jev votes on each note, your code files it, you review the low-confidence pile.
-
jevbench
Preregistered, bias-corrected test of TypeSafe Jev's calibration under human disagreement (ChaosNLI, 100 labels per item)upstream description
-
padflow-jev-evals
Typed-decision benchmark from PadFlow (land development SaaS): schemas, anonymized labeled rows, and a runner for confidence-calibrated models like TypeSafe Jev.upstream description
-
assay-001
ASSAY-001: independent, pre-registered verification of TypeSafe Jev's calibration and type-safety claims. Split verdict, published in full.upstream description
-
Example: confidence-gated escalation
Routing with an act-or-escalate gate, where the policy function is deliberately left unimplemented because the thresholds are yours to choose.
-
jev-asks-until-sure
A twenty-questions guesser that keeps asking until Jev's calibrated confidence crosses a threshold — or gives up and says soupstream description
-
n8n-nodes-typesafe-ai
n8n community node for the TypeSafe AI System One API — typed yes/no, choice and score questions with calibrated probabilitiesupstream description
-
Probing Jev's behaviour with repeated API calls
Independent Korean-language notes reporting that reversing the order of options shifted a probability enough to flip a 0.9 threshold.
-
An early-access test of TypeSafe's Jev: calibrated judgments for half a cent
The best independent test found: 24 Norwegian documents on one pinned model version, opening with a case the model got wrong while correctly reporting low confidence.