User repository on GitHub
Fast structured Android control loops with TypeSafe Jev and Mobile MCP
- Forks
- 4
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Jev Mobile — Friedjof
Experimental Android sub-agent that gives Jev a bounded list of technically valid UI actions per step, with requirement-driven subgoals, durable SQLite tasks, confidence gates and escalation instead of coordinates.
- Stars
- 10
- Language
- Python
- License
- MIT
- Reading time
- 3 min
- Published by
- Community
- Author
- Friedjof
- Last verified
- Sep 2026
About this project
An autonomous, durable Android sub-agent with TypeSafe Jev as its System-One decision provider. jev-mobile accepts a high-level task, persists it in SQLite, and lets one durable worker execute observe → normalize → decide → mutate → verify against a USB-connected Android device. Jev receives concise semantics and technically valid actions; it never generates coordinates, MCP calls, or arbitrary code. Traditional mobile-agent stacks repeatedly involve a large model in every UI step:
Quoted from the project's README. Read the full README.
Highlights
- Jev chooses only among technically valid actions; it never generates coordinates, MCP calls or arbitrary code.
- Durable Core V1 was hardware-validated for notes and checklists, text input, crash recovery, persistence verification and foreign-editor protection.
- Tasks persist in SQLite with leases, checkpoints, cancellation and worker recovery, and mutations are journaled before execution.
- Low confidence, a small top-two margin, repeated states, loops, timeouts and step limits stop automation.
- Exposes six MCP tools over stdio or Streamable HTTP: start_task, get_task, get_task_events, cancel_task, answer_task and get_device_status.
Quickstart
bash
uv run jev-mobile worker --backend portal-adb --serial YOUR_ANDROID_SERIAL
uv run jev-mobile task start "Create a Google Keep checklist titled Shopping with Eier, Brot and Milch"Watch out
MIT-licensed. Requires Python 3.12+, uv, an Android device with USB debugging and a TypeSafe API key; payments and account changes are out of scope and broad app coverage is unsolved.
Questions about this entry
Short answers drawn from the linked project or page.
No, and the scope is deliberately narrow. Durable Core V1 was hardware-validated for notes and checklists, text input, crash recovery, persistence verification and foreign-entity write protection, and it stops rather than clearing unrelated application state. Payments, account changes and other sensitive operations are out of scope without explicit policy and approval work.
Neither. Portal ADB is a local USB-only adapter, and the companion bridge AccessibilityService binds to loopback on the phone while the Python process opens a temporary adb forward tunnel, so no LAN port is exposed and no cloud relay is used. You still have to enable the service yourself in Android settings.
Not by construction. Each step hands it a bounded catalog of technically valid actions such as create entity, set title in title field, verify persisted result, and the worker maps the chosen id to the backend through the DeviceAdapter. Invalid actions, stale references and invented coordinates are rejected before anything reaches the device.
Mutations are journaled before execution and reconciled afterwards, tasks live in SQLite with leases and checkpoints, and worker recovery resumes from the checkpoint. Uncertainty escalates instead of guessing: low confidence, a small top-two margin, repeated states, loops, timeouts and step limits all stop automation, and retry is refused once any Android mutation has begun.
Six MCP tools over stdio or Streamable HTTP: start_task, get_task, get_task_events, cancel_task, answer_task and get_device_status. Queue listing and safety-checked retry stay operator CLI commands on purpose, so the public MCP contract is exactly those six and nothing device-level is exposed.
Providers receive the same goal, a compact semantic state and a fixed list of candidate actions, never raw coordinates or arbitrary code. Credentials go in .env, and TYPESAFE_API_KEY_FILE is preferred for services: it takes precedence over the direct variable, is trimmed when read, and fails closed if it is missing, empty or unreadable.
Yes, but only as a planner. --plan-first asks an OpenAI-compatible LLM for bounded steps and visible completion criteria, and --plan-on-escalation asks for one short text-only recovery plan, configured through LLM_BASE_URL, LLM_API_KEY and LLM_MODEL. The planner holds no device tools, cannot create actions, cannot bypass candidate-action or risk checks, and is attempted once by default.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Jev in front of the HeyGen MCP
X: Playwright CLI + Jev vs Playwright MCP
X: A local MCP server that cuts computer use from 23s to 0.8s. Cartwright built Cortex, a local MCP server where…
X: A side panel that drives any site's WebMCP tools
Other listings of this project
Forks and copies of this project in the directory. This entry is the canonical one.
An Android automation agent that decides from the accessibility tree instead of screenshots: one TypeSafe Jev request per step picks the operation and target, ADB executes, and an optional vision fallback takes over when the fast path gets stuck.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
xinwang-nwpu/jev-mobilePython
xinwang-nwpu/jev-mobilePythonMore Jev resources like this
An MCP server and agent skill that verifies coding-agent work by driving the real running app and returning pass, fail, or couldn't-tell verdicts with the file:line to fix; its listing frames the approach as Jev-style machine-native runtime perception.
Jev Tools & CLIs#community#typescript#mcp
Community
reticlehq/reticleTypeScript
reticlehq/reticleTypeScriptA Python CLI that connects multiple robots under one contract with an optional Jev stepper: Jev picks among permitted discrete calls while an LLM writes poses and prose and the executor enforces safety gates.
Jev Tools & CLIs#community#python#robotics
Community
rokbenko/quackdPython
rokbenko/quackdPythonA self-hosted, versioned skills library for AI agents with HTTP MCP, scoped client keys, and opt-in Jev recommendations that rank skills for a task using your own OpenRouter, TypeSafe, or Vercel key.
Jev Tools & CLIs#community#typescript#mcp
Community
kitze/skillboxTypeScript
kitze/skillboxTypeScriptAn MCP server that puts Jev in the coding loop for Cursor, Codex, and other MCP clients: coding-loop checks, ranking, verification, and review tools.
Jev Tools & CLIs#community#mcp#coding-agent
Community
burnigtm/jev-mcpTypeScript
burnigtm/jev-mcpTypeScriptPython MCP server that exposes classify, score, check, match, and screen tools to MCP-compatible agents, returning typed judgments over a closed answer set with token usage and latency. Ships generic triage and routing question examples.
Jev Tools & CLIs#community#python#mcp
Community
blakestone-x/jev-mcpPython
blakestone-x/jev-mcpPythonAn agent skill for deciding where typed judgments belong beside code, policy, and generation, with Jev examples for question design and abstention plus an offline probability and threshold evaluator.
Jev Tools & CLIs#community#skills#agents
Community
24601/AugustusPython
24601/AugustusPythonFrom 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.