User repository on GitHub
High performance coding agent harness written in rust
- Forks
- 2.4k
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.Homepage
CommunityJev Tools & CLIs
Jcode
Rust coding-agent harness (MIT, ~20k stars) whose crates/jcode-base/src/jev.rs is a typed TypeSafe Jev decision transport, used for memory pruning (memory_jev.rs), browser navigation (browser_jev.rs), and voice routing (voice_intent.rs) with separate provider selectors per purpose.
- Stars
- 20.4k
- Language
- Rust
- License
- MIT
- Published by
- Community
- Last verified
- Oct 2026
About this project
The most RAM efficient harness The most intelligent harness. Website · Docs · SDK · Benchmarks · Features · Install · Quick Start · Further Reading · Contributing. Need Homebrew, source builds, provider setup, or want an agent to set it up for you? Jump to detailed installation.
Quoted from the project's README. Read the full README.
Used inJev for Software & DevTools
- Languages
- Rust
- Platforms
- CLIOpenRouter
- Capabilities
- Open sourceRuns locallyRequires API key
Highlights
- RAM benchmarks in the README put one jcode session at 27.8 MB PSS with local embeddings off, versus 386.6 MB for Claude Code.
- The typed Jev transport defines separate provider environment variables: JCODE_MEMORY_JEV_PROVIDER, JCODE_BROWSER_JEV_PROVIDER, and JCODE_VOICE_JEV_PROVIDER.
- Requests cap at 24 questions (64 for TypeSafe direct); the voice path hedges at 300/700/1500 ms because TypeSafe latency is bimodal per connection.
- Headless swarm workers add roughly 3.1 MB PSS per extra session in the project's Linux measurements (jcode v0.89.19-dev).
Watch out
MIT license; Jev usage requires provider credentials (TypeSafe direct or an OpenRouter-compatible gateway), and the quoted benchmark numbers are self-reported in the README.
Questions about this entry
Short answers drawn from the linked project or page.
A typed decision transport in crates/jcode-base/src/jev.rs serves three purposes: memory retrieval and pruning (memory_jev.rs), browser navigation (browser_jev.rs), and voice interaction routing (voice_intent.rs), each with its own provider selector.
The README's Linux measurement (jcode v0.89.19-dev) reports 27.8 MB PSS for one session without local embeddings and about 3.1 MB more per additional headless session, compared with roughly 164 MB per extra Claude Code session.
The typed transport caps requests at 24 questions, or 64 when talking to TypeSafe directly, and each purpose selects its own provider through environment variables.
Because TypeSafe latency is bimodal per connection, the voice path hedges at 300, 700, and 1500 ms rather than waiting on a single timeout.
Separate ones per purpose: JCODE_MEMORY_JEV_PROVIDER for memory pruning, JCODE_BROWSER_JEV_PROVIDER for browser navigation, and JCODE_VOICE_JEV_PROVIDER for voice routing.
On macOS and Linux run curl -fsSL https://jcode.sh/install | bash; on Windows 11 with PowerShell 5.1+ run irm https://jcode.sh/install.ps1 | iex. Homebrew and source builds are also documented.
MIT. Using Jev needs provider credentials, either TypeSafe direct or an OpenRouter-compatible gateway, and the quoted memory benchmark numbers are self-reported in the README.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
Hacker News: jevmem keeps Claude Code project memory by asking Jev which turns are worth saving.
avinashjetwani61 pointsJevmem – automatic project memory for Claude Code, built on JevRead the thread on Hacker NewsX: A usage router for Grok Bot
X: The compaction plugin, in Chinese
X: A prompt box that fills itself in
More Jev resources like this
3.4kGitHub stars
Vercel Labs' Unix-like coding agent CLI written in Zig, shipping a typesafe_permission_reviewer builtin that runs the agent's permission decisions through Jev rather than an LLM call. Model-agnostic and embeddable via libfx and ACP.
Jev Tools & CLIs#community#coding-agents#cli
Community
vercel-labs/fxZig
vercel-labs/fxZig21GitHub stars
MIT-licensed Python tool router for coding agents that keeps obvious calls on a fast path, asks Jev (default typesafe/jev-1.13 via OpenRouter) for genuinely ambiguous choices, and can run bounded MCTS when multi-step consequences matter. Ships a CLI, MCP server and hook integrations.
Jev Tools & CLIs#community#routing#mcp
2.6kGitHub stars
A CLI that asks Jev which files and declarations answer a repository question, then prints verbatim excerpts for a coding agent to use.
Jev Tools & CLIs#community#typescript#cli
Community
dzhng/jevgrepTypeScript
dzhng/jevgrepTypeScript634GitHub stars
A CLI browser automation tool with optional Jev semantic mode: Surf observes, Jev selects the next action, Surf validates and acts, Jev checks the goal.
Jev Tools & CLIs#community#browser#cli
Community
nicobailon/surf-cliTypeScript
nicobailon/surf-cliTypeScript209GitHub stars
A Python harness for System One models. It runs Jev and other decision models locally, or through HarnessRouter.ai.
Jev Tools & CLIs#community#python#open-models
Community
HarnessRouter/SystemOneHarnessPython
HarnessRouter/SystemOneHarnessPython159GitHub stars
Rust agent workbench for ongoing work that keeps branch decisions in their own crate and backend notes in the adapter. Its first decision backend is Cloudflare's Clef, borrowing Clef's System One-shaped state and questions request instead of reinventing the interface.
Jev Integrations & Adapters#community#agent-runtime#rust
Community
holon-run/holonRust
holon-run/holonRustFrom the Jev guides
Original write-ups on what this entry is for.
The fastest-growing corner of the Jev ecosystem is coding agents. A tour of the MCP servers, tool-call gates, context pruners, and skill rankers built for that loop.
Jev Tools & CLIs#agents#mcp#coding-agent
Read article
The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.
Jev Tools & CLIs#gateway#vercel#cloudflare
Read article
Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.
Jev Tools & CLIs#open-models#local#onnx
Read article
From the Jev community
Posts from builders shipping with Jev right now.