Read & learn
Written guides.
Understand Jev, one idea at a time. Walkthroughs, recipes, and writeups — our notes first, the original next.
Explore the Jev ecosystem.
Learn, build, and play with Jev — from deep technical guides to creative experiments.
Watch & learn
Video guides.
See an idea take shape. Tutorials, demos, and deep dives, organized by topic and credited to their creators.
Make & explore
Experiments.
See what builders made with Jev. Games, tools, repositories, articles, and more — traced to their sources.
The directory03 / 03 · Experiments
Find something worth exploring.Experiments
Every category, from games and tools to repositories and writeups, traced to a primary source.
Same ideas.
Different paths.
This is the full registry, not just the featured picks. Figures like speed, cost, and stars are maker-reported or captured snapshots, not JevMade measurements.
311 experiments · showing 241–300
An OpenAI- and Anthropic-compatible gateway where a small System One model picks which larger model should answer each prompt.
SOURCE SCREENSHOTEXPAND ↗
from the source
const { json, ms } = await post(DECISIONS_URL, { model: JEV_MODEL, state: { request: c.p
An MCP server and agent skill for using System One at design time: decompose a judgment into questions, lint them, calibrate thresholds on labelled rows and rerank.
SOURCE SCREENSHOTEXPAND ↗
from the source
const result = await gateway.systemOne(args.state, args.questions, args.model, extra?.si
A local agent harness that puts Jev in front of an ordinary LLM and pays for the LLM only when the work needs prose.
SOURCE SCREENSHOTEXPAND ↗
from the source
payload = {"model": self.model, "state": state, "questions": questions}
A coding-agent CLI that splits the work: a decision model picks the next tool, scores progress and judges whether the goal is reached, while an LLM only fills in arguments and writes code.
SOURCE SCREENSHOTEXPAND ↗
from the source
const body = JSON.stringify({ state, model: this.config.model, questions })
Gates agent tool calls through detectors, Jev judgments, and deterministic policy before allowing, blocking, or requesting approval.
SOURCE SCREENSHOTEXPAND ↗
Reads completed Claude Code or Codex sessions and assigns typed judgements to each turn for later analysis.
SOURCE SCREENSHOTEXPAND ↗
Put many MCP servers behind two tools while Jev discovers the best enabled tool and the calling agent supplies its schema-bound arguments.
SOURCE SCREENSHOTEXPAND ↗
A Rust CLI turns Jev judgments into stable JSON, resumable batches, assertions, and meaningful shell exit codes.
SOURCE SCREENSHOTEXPAND ↗
A small CLI ranks local Agent Skills against a request, then returns suggestions without invoking anything automatically.
SOURCE SCREENSHOTEXPAND ↗
Add three Claude Code hooks that flag dangerous commands, failed tool calls, and incomplete work with Jev.
SOURCE SCREENSHOTEXPAND ↗
Screen fetched pages and files for prompt injection before coding agents load them, blocking content that crosses configured thresholds.
SOURCE SCREENSHOTEXPAND ↗
Check each coding-agent edit against a team style guide and send violations back into the session.
SOURCE SCREENSHOTEXPAND ↗
Riley Brown put Jev behind an agent’s model router to choose which model handles each task.
ACTUAL RECORDING12 SEC ↗
ImRobot’s hook-based gate blocks dangerous agent commands until a person approves that single execution.
SOURCE SCREENSHOTEXPAND ↗
Milind S added Jev to OpenMausBot to choose both the teammate and model for each task.
ACTUAL RECORDING12 SEC ↗
Cyril split agent work so Claude Code writes while Jev handles repeated decisions on suitable tasks.
SOURCE SCREENSHOTEXPAND ↗
Jetwani Avinash’s Claude Code gate asks Jev whether each message deserves one line in a repository memory file.
ACTUAL RECORDING12 SEC ↗
Paulius used Jev to launch agents on Clonk’s visual canvas without waiting for a full LLM loop.
ACTUAL RECORDING12 SEC ↗
Kelbie built a Jev tool that scores every code chunk for relevance to a developer’s current question.
ACTUAL RECORDING12 SEC ↗
Cortex is a local MCP server where Claude plans tasks and a faster action layer handles clicks across browsers and Mac apps.
SOURCE SCREENSHOTEXPAND ↗
Oskar is building an open-source framework that uses agents and Jev to run end-to-end tests across web and mobile.
ACTUAL RECORDING12 SEC ↗
grok-jev-reflex is an open-source router where Grok plans and acts while Jev handles cheaper yes-or-no decisions between steps.
SOURCE SCREENSHOTEXPAND ↗
Ishwar demonstrates instant Jev compaction during a long Claude Code server-redeployment session.
SOURCE SCREENSHOTEXPAND ↗
Action Gate asks Jev to score risky agent actions and routes a simulated $50,000 transfer to human review.
ACTUAL RECORDING12 SEC ↗
jev-studio packages Jev’s three primitives as MCP tools and a CLI with dry-run provenance.
SOURCE SCREENSHOTEXPAND ↗
Monid connects Jev and OpenRouter to a catalogue of tools for lead scoring, research, content analysis, and site audits.
SOURCE SCREENSHOTEXPAND ↗
Kiyoro used Jev to detect that Opus 5.5’s default effort had changed from high to medium.
SOURCE SCREENSHOTEXPAND ↗
Sawyer Hood uses Jev to choose an agent, model, computer and working folder from a single prompt.
ACTUAL RECORDING12 SEC ↗
Neha Sharma built a real-time demo that uses Jev to route work among AI agents.
SOURCE SCREENSHOTEXPAND ↗
erKam built a Jev check that quotes violated project rules back to Claude Code so it can rewrite its response or edit.
SOURCE SCREENSHOTEXPAND ↗
Jev chooses a skill, tool, and parameters before a Slack agent acts, which its maker says doubled its speed.
ACTUAL RECORDING12 SEC ↗
Jev directs a coding agent toward relevant files and flags possible test gaps, while deterministic tests verify the result.
SOURCE SCREENSHOTEXPAND ↗
Read browser-agent activity logs every three seconds and average the resulting risk score across the session.
SOURCE SCREENSHOTEXPAND ↗
OpenCodex asks Jev to choose a model and reasoning level for each Codex request, using the proxy's configured route whenever Jev is unavailable or returns an invalid answer.
SOURCE SCREENSHOTEXPAND ↗
from the source
requestBody = JSON.stringify({ model: JEV_MODEL, state, questions: buildJevRouteQuestion(options.candidates),});
Docker Agent asks Jev whether a pending tool call is clearly read-only or needs the user's approval, treating uncertain calls cautiously.
SOURCE SCREENSHOTEXPAND ↗
from the source
type: choicechoices: read_only: Clearly read-only, with no credential access or external data transfer. risky: Modifies or deletes data, executes opaque code, accesses credentials, or transf... unknown: Insufficient information to establish the operation's effects.
Agent Native asks Jev which optional tools an agent should see at the start of a request, based on the user's recent context.
SOURCE SCREENSHOTEXPAND ↗
from the source
candidateStateKey: "candidate_tools",answerKey: "best_tool",question:Which tools should be loaded into the agent context first for this task? Pick the most u...limit: prefetchLimit,
Memorax asks Jev whether searching saved coding memories would materially help with the current prompt and, when available, the previous turn.
SOURCE SCREENSHOTEXPAND ↗
from the source
body: JSON.stringify({ model: JEV_MODEL, state, questions: { search_needed: SEARCH_QUEST...const answer = isRecord(body) && isRecord(body.answers) ? body.answers.search_needed : u...const probability = answer.noul;return { ok: true, decision: probability >= 0.5 ? "search" : "skip", probability, model:...
Interactive Shell uses Jev to decide when terminal work needs attention and which pre-authorized action may help.
SOURCE SCREENSHOTEXPAND ↗
from the source
questions.attention = { type: "choice", instructions: `Choose the primary visible state using only explicit current evidence..... criteria: { working: "routine work is active, retrying, progressing, or healthily waiting...",
Kimaki's OpenCode auto-mode asks Jev whether a pending tool action may run without another prompt, using the action, its arguments, and the latest user message.
SOURCE SCREENSHOTEXPAND ↗
from the source
state: serializedState,questions: { allow: { type: 'boolean', instructions: `Should this pending OpenCode tool action run automatically? ${CLASSIF...
ctx sift uses Jev to keep useful or contradictory passages while shrinking an agent's tool output.
SOURCE SCREENSHOTEXPAND ↗
from the source
let judgment = client.select(selection.task(), selection.candidates(text));if let Some(scores) = judgment.scores.as_deref() && let Some(proposal) = selection.propose(scores){ selected_count = proposal.kept_count();
mem-jev asks Jev to judge ambiguous agent memories, then folds five semantic signals back into retrieval ranking.
SOURCE SCREENSHOTEXPAND ↗
from the source
evaluation, err := service.evaluator.Evaluate(ctx, request.State)record, err := NewJudgmentRecord(...)return ServiceResult{ Disposition: DispositionCommitted, Features: record.Features,
Keel can ask Jev which installed coding provider and model should receive a new task.
SOURCE SCREENSHOTEXPAND ↗
from the source
let input = SelectionInput { state: DecisionState { task, context, state_version }, candidates: prepared_routes,};let outcome = selector.select(input, &DecisionBudget::new(1)).await;
Record live Jev decisions as redacted cassettes, then replay or perturb them without another provider call.
SOURCE SCREENSHOTEXPAND ↗
from the source
JevResponse response = delegate.send(request);repository.save(new Cassette(...));
Add Jev-backed tool approval, chat triage, transcript curation, and model routing to OpenClaw.
SOURCE SCREENSHOTEXPAND ↗
from the source
const scored = await this.client.score({if (scored.level >= this.approvalLevel) {
DLQ Inspector asks Jev whether an ambiguous dead-letter message should be replayed, fixed, avoided, or investigated.
SOURCE SCREENSHOTEXPAND ↗
from the source
resp, err = a.Client.Evaluate(ctx, state, jev.ClassificationQuestions())out.Classification = jevChoiceToClassification(verdict.Classification)
Any Auto asks Jev whether coding-agent tool calls are risky, authorized, and policy-compliant before approving them.
SOURCE SCREENSHOTEXPAND ↗
from the source
json!({"model":self.config.approver.model,"state":input.state,"questions":questions})let mut result = into_review(&response)?;
Supercov asks named quality and security questions about source files, then turns Jev's answers into inspectable findings.
SOURCE SCREENSHOTEXPAND ↗
from the source
if let Some(health) = catalog::health(values) {weighted.push((answer.bytes, health));
jevbrief fixes the least glamorous part of a Jev integration: what the model actually gets to read. Adapters for CI runs, pull requests, OpenTelemetry exports, and other noisy sources cut state down to what matters, and every dropped line gets a reason.
SOURCE SCREENSHOTEXPAND ↗
An ACP and MCP adapter that hands Jev's three typed question types to any LLM agent, built for Roblox computer use.
SOURCE SCREENSHOTEXPAND ↗
When the robot acts up, Jev plays detective: it chooses which diagnostic tool to run next, while the tools themselves stay plain deterministic code.
SOURCE SCREENSHOTEXPAND ↗
Points a debugger at your bug and lets Jev decide every next step, so the expensive model only wakes up for the hypothesis and the fix.
SOURCE SCREENSHOTEXPAND ↗
Keeps a record of what you and your coding agent agreed to build, in plain Git commits where the next session can find it.
SOURCE SCREENSHOTEXPAND ↗
Watches what you type into Claude Code and decides who should handle it: your slash commands go straight through, everything else gets checked against your rules.
SOURCE SCREENSHOTEXPAND ↗
Semantic lint rules written as plain-English yes-or-no questions, with Jev returning a calibrated probability instead of free text.
SOURCE SCREENSHOTEXPAND ↗
One Jev call before every commit checks that your message actually describes the diff, and flags debug leftovers and pasted secrets while it is at it.
SOURCE SCREENSHOTEXPAND ↗
Gives the Pi coding agent a Jev-powered tool for the small stuff: classify this report, triage those issues, score these candidates, all in one batched call.
SOURCE SCREENSHOTEXPAND ↗
A Python library that reranks and relevance-filters RAG retrieval with Jev, scoring documents for how much they help answer the query.
SOURCE SCREENSHOTEXPAND ↗
A Go reverse proxy that sits between nginx and your LLM backend, judging every request's user input with Jev and blocking harmful traffic transparently.
SOURCE SCREENSHOTEXPAND ↗
A semantic code linter that judges your source against configurable plain-language rules and prints per-finding confidence.
SOURCE SCREENSHOTEXPAND ↗
A chaperone for agents doing web research. Nothing gets fetched or trusted without Jev weighing in on it first.
SOURCE SCREENSHOTEXPAND ↗
Jev, briefly
A decision model, not a chatbot.
A chat model generates text token by token and hopes you parse it. Jev never generates a word. Your code sends the state of the world plus typed questions; Jev returns every answer in one parallel pass — typed values with calibrated probabilities and a confidence score your code can trust. About 70–500 ms end to end, $0.042 per million input tokens, output free. Trained with what TypeSafe calls Reinforcement Learning for Calibrated Decisions.
choice Pick one of these.
You define the options. Jev returns the chosen option, a probability for every option, and a confidence. The workhorse of routing, agents, and games — the answer space is always legal, so there is nothing to parse and nothing to hallucinate.
score Rate this on a rubric.
You describe ordered levels — say trivial / normal / critical. Jev returns the level, the probability of each, and a confidence. Scores turn fuzzy judgment (“how severe is this log line?”) into a number-free decision your code can branch on.
noul Is this statement true?
You assert a statement; Jev returns the probability that it is true — a noul. Moderation, verification, guardrails, “does this diff actually fix the bug?”: one question, one calibrated probability.
Official material lives at typesafe.ai and docs.typesafe.ai. JevMade is an independent community registry — not affiliated with or endorsed by TypeSafe AI.
Recurring lessons
Patterns that keep showing up.
One decision per tick
Games, drones, trading bots, and browser agents all converge on the same loop: serialize the state, ask one decisive question, act, repeat. Jev’s latency makes the loop feel instant — the model lives inside the control loop, not outside it.
Confidence as a gate
The answer says what; the confidence says whether to act. The most reliable entries threshold on confidence to route edge cases to a slower model or a human — automation with an honest escape hatch.
Shrink the choice space in code
Makers rarely ask Jev to pick from everything. Local tactics prune 225 gomoku moves to ~40; DOM filters turn a page into an element table; code narrows options, Jev judges within them.
Jev judges, LLMs talk
A recurring split: Jev makes every decision cheaply and instantly, and a small LLM is only invoked when a human-facing string must actually be written. Decision and generation are separate budgets.
Made something with Jev?
Send the link — repo, live demo, post, or video — plus a line on what it does and which primitives it uses. Every entry is verified against its primary source before it ships.
Submit your experimentSign in first so we can reply to you. Submissions are reviewed, not published automatically.