Jev AgentA source-verified directory of 487 Jev projects — browser agents, model routers, SDKs and MCP servers — each one showing the exact decision it hands to Jev. Curated by the Awesome Jev community, mirrored here under MIT.
Give Jev the current state and a few options. Get the Agent’s next step.
Core call · setup in the full example
const nextStep = choice("What next?", { login: null, askUser: null, retry: null });
const { answers } = await client.systemOne({ state: "Login form is ready", questions: { nextStep } });
console.log(answers.nextStep.choice);Node.js 20+ · Set TYPESAFE_API_KEY on your server. This page shows code and makes no model requests.
An independent supervisor loop that reads worker diffs, logs and tests, asks Jev Nouls about stuck / off-track / verify, then applies a Python policy.
AsyncTypeSafeClient.system_one with default jev-latest sends Noul questions for supervision.
Supervision does not write the worker's code. Distinct from cataloged Shifty-Eye-Games/foreman-jev.
OrchestKit can optionally use Jev to classify coding sessions and set their colors when confidence meets a threshold.
Classifies the first task prompt and branch state by work type; local policy accepts the result or falls back.
Distinguishes sessions by work type and supports a shadow comparison mode.
1. Evaluates the configured dataset and measures uncertainty. 2. Selects ambiguous rows plus a random audit sample for you to label. 3. Uses your accumulated labels and optional rationales to run GEPA. 4. Shows the score, certainty change, and proposed definition diff. 5. Lets you accept, reject, rewind, or resume later.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Fish-style Zsh history suggestion tool ranked by Jev, ordering candidate commands from local history based on context.
Packages current command input and historical candidates into a Jev request, ranking the most plausible command for inline suggestion.
Combines the safety of local shell history with Jev semantic relevance, filtering out stale or risky command candidates.
A Python CLI and Codex Skill that recommends a suitable installed skill for the current task using TypeSafe Jev Choice and Noul checks without executing candidate skills.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
A Python CLI for semantic predicates, routing, scoring and filtering over text or JSONL.
Sends input to Jev and applies local probability thresholds to outputs and exit codes.
Adds typed decisions to Bash and CI with explicit output and failure states.
An experimental coding CLI where Jev selects AST productions for Python or Bash, with a separate command-line decision mode.
Chooses from bounded syntax and action options; local code constructs programs or invokes tools.
Shows coding choices and command-line judgments in inspectable traces.
The jevctl command-line tool connects verification, classification and scoring questions to text input and scripts.
Sends input and bounded answer options to Jev, then reports judgments and probabilities using local thresholds.
Provides structured results for pipelines and CI, with request inspection and dry-run support.
A CLI for Jev pick, rate, check, rank, triage, and guard tasks, with optional hooks before Agent tool calls.
Converts input state and options into structured questions, returning results or risk scores for local policy.
Reuses the same decision commands in scripts and Agent workflows.
Filters text, structured records, functions, and diff hunks against plain-English descriptions using Jev Noul judgments.
Jev judges whether each input unit matches the user description; local code applies the probability threshold and returns matching source material.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Ranks text along a plain-English criterion using pairwise Jev Noul comparisons and a locally fitted Bradley-Terry scale.
Jev judges whether text A ranks higher than text B on the supplied criterion; local code schedules comparisons and fits a Bradley-Terry scale with standard errors.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
grep for what code does, not what it's called. Semantic code search powered by TypeSafe Jev.
Measured (2026-09-19, jev-1.13.0): a 896-chunk TypeScript `src/` tree in 1.8 s for $0.010 (240k input tokens); repeat query 0 s from cache.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Software development framework for coding agents enhanced with Jev for dependency vetting, completion gates, and debugging decisions.
Invokes Jev at workflow checkpoints to evaluate change validity, test sufficiency, and package safety before advancing agent state.
Constrains coding agent trajectories using fast discrete judgments, preventing hallucinated dependencies and premature task completion.
A CLI and stdio MCP server for typed Jev judgments over text or JSON.
Submits noul, choice and score questions and emits JSON or a primary value.
Accepts files and stdin for use in shell scripts and MCP clients.
An optional TypeSafe decision client in the Rust coding terminal Rift for bounded Jev judgments.
Sends state and typed questions to System One and parses answers for terminal workflows.
Adds a separate decision interface alongside generative coding models.
Rule-based CLI and agent skill that evaluates text and markdown files against custom rulesets for AI slop, grammar, and technical doc quality using Jev scores and line-by-line violation flags, then guides an AI agent to auto-fix violations.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
JEV never writes a review or changes a specification. It returns typed decisions and probabilities; deterministic Python code decides whether to pass, request review, or block.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Don't burn your expensive main model on grep-and-guess grunt work — let jev rank the whole repo, and save the main model for reading the right files and writing the right code.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Use it to assess unfamiliar instrumentation, review candidates for reduced retention, and selectively filter metrics before they reach a primary backend. Inference runs asynchronously, and cached assessments let subsequent batches use the same decision without another API call.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
I kept watching coding agents burn context on decisions that aren't hard - triage 400 tickets, tag 600 files, route to one of six teams. jev-mode moves those verdicts to a typed-judgment model. I A/B'd it: 78% fewer tokens, 16x less work-attributable input, accuracy 96.1% vs 93.7%. Python, no deps, MIT.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
A CLI for asking Jev classification, yes/no and scoring questions over text or JSON input.
Evaluates typed questions over one input and returns choices and probabilities as JSON.
Accepts standard input and supports direct TypeSafe calls or a Vercel gateway.
Ultra-fast, fail-open advisory decisions and verbatim extractive reading view for AI coding agents and CLI pipelines
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
**Jev scores 77.0%; Qwen3.5 2B with a 2,048-token thinking budget scores 82.0%; Qwen3.5 4B with the same budget scores 96.7%.** This small benchmark makes Jev's observed reasoning limits tangible through nine local-model settings.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
A GitHub Action that compiles `.jev.yml` rules into Jev questions and fails, comments, or passes from the numbers.
Rules become at most twelve questions (`MAX_JEV_QUESTIONS` in src/types.ts). AKIA and private-key armor from src/redact.ts fail locally in src/pipeline.ts with no Jev call.
Does not write reviews, suggest patches, or approve. Policy is read from the base-ref config.
A community-maintained directory and radar for Jev, highlighting open-source projects with verified code and clear decision architectures.
Jev is TypeSafe's decision model. This directory groups projects by how they use choices, scores and probability judgements, with implementation links to help you assess fit for your own task.
TypeSafe ↗We prioritize open-source projects with clear Jev implementation evidence. Repositories without verifiable source code are excluded from the main directory.
Directory data and design from Awesome Jev by logicrw, used under the MIT License.
