Awesome Jev

Jev is TypeSafe's System One model for probability-based yes/no judgments, choices, and scores rather than generated text.

Check permissions and data handling before running a project. Source review is not a runtime test or a security audit.

• Getting Started
• Demos
• SDKs and Skills
• Browser and Desktop Tools
• Agent Tools
• Apps and Integrations
• Games and Creative Projects
• Independent Models
• Reading and Research
• Supporting Drivers

Start with the quick-start example, then build a typed classifier, use Jev in an agent, or explore a local alternative. The source notes record setup, privacy, and evidence limits for the selections below.

A few examples to watch before digging into the code. These are creator recordings and publisher previews, not independent benchmarks. Linked previews open the original recording or post.

• Jev Ultrafast - A flight-search recording where Jev chooses browser actions and elements; a separate model writes field text. Browser Use · @gregpr07.

• Jev Voice Browser - Partial speech becomes action and target choices, then Playwright operates the page. Moritz Kremb · @moritzkremb.

• jev-align - Label uncertain examples and review proposed classifier or scoring-rubric changes. Sutro. This edits definitions, not model weights.

• TypeSafe Computer Use - A Mac desktop recording where OCR and accessibility observations become action choices. @awlevin. Real use controls the mouse and keyboard and needs desktop permissions.

• Jev Browser Skill - A Hacker News walkthrough showing observed controls, bounded choices, and local execution in a small reference implementation. zurfyx.

• tisco - Search video transcripts, inspect matches, and approve clip organization. cairodavila · screenshot published by OpenRouter. This is a screenshot, not a video; Jev judges transcripts rather than video pixels.

• JevPDF - Find PDF passages by meaning and highlight relevant lines. Kyle McLaren / @kylemclaren. Author-hosted repository animation; meaning search sends extracted text to TypeSafe.

Newly added: plain-English diff checks, memory-graph decisions, JSON record filtering, Cua driver decisions, context pruning, and a desktop Jev workbench. These are newly curated September posts; each entry records source and playback limits.

More to watch: Shapeshift's adaptive cards, a Rubik's cube coach, sentence-level reading highlights, CLI suggestions, site-search reranking, and semantic SQL. Each entry links the creator post and records its source and access limits.

The full creator gallery retains the other videos, images, and attribution. Browse the X index, OpenRouter roundup, or computer-use comparison for more examples and limitations.

• TypeSafe Python SDK - Official Python client with a support-ticket routing example.
• TypeSafe JavaScript SDK - Official JavaScript and TypeScript client.
• TypeSafe Go SDK - Community Go client with typed question builders, response helpers, retries, and runnable examples.
• Official TypeSafe Skill - Request-writing guidance for Noul, Choice, and Score questions.
• gut - Python decision API with Jev, local, and OpenAI-compatible backends, plus batching, caching, and human-escalation thresholds.
• Tenbin - MCP server and skill for linting questions, evaluating labeled examples, and choosing thresholds.
• wellposed - Offline request linter for missing options, broken state references, and question types.
• jevrs - Async Rust client with typed answers, derive macros, and native or WASI transports.
• Jevify - Agent skill for finding places to use Jev and planning comparisons with existing code.
• Jev Logs - Scores log entries to decide which need a closer look.
• Skillbox - Self-hosted skill library with optional Jev-based recommendations.

• Jev Ultrafast - Browser-action and element selection with a separate model for text entry.
• Jev Voice Browser - Voice-controlled Playwright browser that chooses actions from partial speech and page controls.
• TypeSafe Computer Use - Mac automation using local OCR and accessibility data, with Jev choosing the next action.
• Cua Driver + jev-use - Python and TypeScript examples where Jev chooses an action, the driver runs it, and code checks the result.
• Jev Social - Bounded social-research loop using Jev choices and socai in the user's Chrome session.
• Jev Browser Skill - Small reference for building and executing browser-action choices.
• Jev Browser - Playwright browser available as an MCP server, CLI, or library.
• Jev Browser Use - Skill using an existing Codex browser connection, with Jev choosing clicks and scrolls.
• Jev Cua - Mac automation through Cua Driver, with optional hosted Jev workflow recommendations.
• jev-bot - Native Mac controls through a JavaScript MCP session, with optional hosted Jev element selection.

• pi-jev-governor - Routes planning, model tier, tool permission, and completion decisions for the pi coding agent, with deterministic gates and human or fail-closed fallbacks.
• softlint - Checks changed diff hunks against plain-English rules and locates findings on added lines.
• Jev Dreaming - Experimental memory graph where Jev filters text and judges memory relationships while Gemini writes the memories.
• fast-jev-compaction - Experimental history pruning that uses Jev judgments to keep, truncate or drop tool-call pairs.
• jev-align - Improves classifier definitions and scoring rubrics using labels and GEPA-proposed edits.
• Winnow - Claude Code context filter that scores tool-output blocks and keeps hidden text locally recallable.
• Jev-Mem - Research memory system where Jev organizes graph memories and guides retrieval while a separate model writes answers.
• patdown - Semantic linter that turns Markdown rules into checks and maps flagged evidence to source lines.
• Oko - Code-search CLI and MCP server with local candidate retrieval and optional hosted Jev ranking.
• jgrep - Semantic matching over code chunks, diff hunks, and CSV rows, published on npm as jevgrep.
• SemDecide - Unix CLI for predicates, choices, scores, JSONL filtering, and CI-friendly exit codes.
• Leanest - Playwright and Vitest test selector with an optional Jev judge and run-tests fallback on uncertainty.
• Supercov - Coverage, security, and code-quality checks where Jev evaluates source files while test coverage stays local.
• Foreman - Checks coding-agent progress and decides when intervention may be needed.
• jev-belay - Claude Code Stop hook that checks local run evidence before judging an unverified completion claim.
• jev-claude-statusline - Claude Code hooks that estimate task progress with Jev and cache it for a status-line display.
• Jev MCP - Tools for checking claims against evidence, screening content, and ranking candidates.
• TypeSafe MCP - Go MCP server for making Jev requests from coding agents.
• jev_search - Experimental file search that ranks candidates by keyword and checks passages with Jev.

• TypeSafe AI for n8n - Official community node for evaluating or routing workflow items with Choice, Noul, and Score questions.
• pen-find - Finds Pen design frames by sending node names, paths, text, and structure to Jev for semantic ranking.
• JevPDF - Meaning-based PDF search with locally extracted text and Jev-ranked line highlights.
• jev-eyes - Locally extracts OCR text and spatial layout from images into inspectable state, with an optional helper that sends that state to Jev.
• ASIMOV Jev output filter - Optional plain-language filtering of JSON records before deterministic output projection.
• Jev dev - Desktop workbench for typed Jev requests, answer inspection and saved run history, with a separate demo mode.
• jevotron - CLI that uses Jev to flag anomalies in selected fields, with file-format adapters, local previews and cached assessments.
• DocJev - Document classification and packet-boundary detection after local parsing or optional cloud OCR.
• tisco - Video-transcript search with previews of proposed clip moves and renames for approval.
• neo4jev - Neo4j graph exploration using edge choices and goal checks inside a bounded beam search.
• MinusPodJev - Podcast ad-detection adapter that scores transcript segments and assembles ad spans in code.
• llm-typesafe - Jev access from the LLM command line or Python, including stdin, templates, and async calls.
• Home Assistant Jev - Typed decisions for Home Assistant entities and automations.
• nospace - Experimental word-spacing interface where Jev chooses splits and a separate model handles spelling.
• Vibe Domain - Domain-name ranking with a no-key heuristic mode and an optional hosted Jev mode.
• Spliit Cloud - Optional expense-category suggestions where Jev picks a category after local dictionary and history matching miss.

• Chirp - Turns coding-agent completion state into short synthesized notification sounds, with Jev choosing bounded mood and voice parameters.
• 1 Million Emojis - Shared emoji canvas where Jev chooses nearby emoji placements and judges whether to finish a stroke.
• Jev Music Playground - Musical-parameter selection with code-generated notes and MIDI, plus a labeled offline mode.
• JevAI for XMage - Magic: The Gathering bots using Jev alone or alongside XMage search.
• PROMPT FPS - Action selection in a browser-based Doom engine using Freedoom assets.
• Jev Plays Snake - Snake agent where code finds legal moves and Jev picks a direction.
• jev-plays-pokemon-red - Branch-point action selection for a Pokemon Red agent requiring a lawful local game dump.
• TypeSafe Mario - Controller-input selection from emulator state, requiring a lawful local game setup.
• JevPilot - Driving simulation with typed steering and speed choices, not a real-world driving system.
• Jev Chess - Hosted shared chess game with move probabilities; public source code has not been verified.

These projects are not official TypeSafe Jev weights, and compatible outputs do not establish equivalent quality.

• Cloudflare Clef / Clef-flash - Open-weight 27B and 9B multimodal decision models with a Jev-compatible API, runnable locally or through Workers AI; not TypeSafe Jev weights.
• AnyJev - Typed-decision readouts for open models with option-order correction and per-question calibration.
• CUA-S1-FORMS - Independent form-filling specialist and dataset, not a general desktop agent.
• Jevlike - Independent option scorer with training code and game examples.
• OpenJev SGLang - Independent typed-decision server using open models and SGLang.
• Open Medical Jev - Independent frozen-model medical decision router with confidence fusion and conformal candidate sets; not TypeSafe Jev weights.
• WaterSheep - Independent open-weight model for yes/no, choice, score and multi-label questions, with option probabilities; not TypeSafe Jev weights.
• TetraJev - Independent research recipes combining letter and per-option readouts from two frozen local models, with agreement-based decision routing.

• Jev in the Wild - Surveys 2,170 public Jev projects and maps early growth, application domains, and decision-use patterns.
• Jev vs an LLM: the numbers builders published - Puts published Jev and LLM runs of the same job side by side (100,000 posts for $0.67 against 214 for $0.98) and lists the cases where the larger model still won.
• Jev benchmark - Publishes a harness and raw per-item results for author-run Jev comparisons on two small classification tasks.

• Browser Harness - Model-neutral CDP connection used by Jev Ultrafast, not a Jev model.

Suggest a project with a concrete Jev role, implementation evidence, and honest limitations. Read the contribution guide and selection policy first.

The larger discovery catalog, local viewer, and refresh history are separate from this editorial file. The gallery selection notes explain the dated traction checks and editorial ordering; this is not a live popularity leaderboard. Discovery, research, and drafting have used AI assistance; this project does not claim an exclusively human-authored history or admission to the canonical Awesome index. See submission status, sources and credits, reuse terms, and the security policy.