jev-engineering-zh
Chinese translation of 'Jev Engineering for coding agents', design notes on a harness where Jev makes every per-turn decision.
This repository is mostly documentation or write-up rather than runnable code.
Overview
jev-engineering-zh is a complete Chinese translation of a compiled working-notes document on building coding agents around Jev, based on design notes attributed to TypeSafe founder Diogo Almeida. It keeps the original structure and all seven figures, and the repository also includes the author's raw source notes so readers can compare notes and finished text. It states that it is an independent compilation, not affiliated with or endorsed by TypeSafe.
The notes start from the question of how you would design a coding agent if language models had no KV cache, and identify six inherited design choices: lossy routing, tool calls crowding context, blind compaction, rarely used sub-agents, restarts that discard state, and battery disputes. The proposed harness keeps explicit, typed, addressable state blocks and asks Jev each turn which blocks to show, whether to reuse the cache, which model and tools to use, and whether a command may run. It also walks through routing arithmetic, token economics and routing by file sensitivity.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Should this context block be shown this turn? | — |
| Choice | Which model and tool should handle this turn? | — |
| Yes / no | May this command run? | — |
How it works
- Store agent state as typed, addressable blocks
- Each turn, ask Jev which blocks to show and whether to reuse cache
- Ask Jev which model, tool and permission apply
- Run the frontier model, sub-agent or cheap model under safety policy
Strengths
- Chinese access to a detailed harness design
- Includes raw notes for comparison
Caveats
- Design notes, not working code
- Independent compilation; not endorsed by TypeSafe
At a glance
- Type
- Project
- Status
- discussion
- Has code
- no
- Has tests
- no
Use cases and integrations
How curated lists describe it
《Jev 工程学：为 coding agent 而作》完整中文翻译 — 保留原结构与 7 张插图.
hellogumbo/awesome-jev
Frequently asked questions
What is this document?
A Chinese translation of working notes on designing a coding agent harness around Jev.
What decisions does Jev make in the design?
Which context blocks to show, cache reuse, model and tool choice, and whether a command can run.
Is it official?
No, it is an independent compilation.
Related projects
Korean course code for attaching Jev to Claude Code, Codex and OpenClaw: skills, hooks, routers and open-model comparisons.
ProjectproductionJavaScriptCompanion code for the ZazenCodes course on decision models: primitives notebook, use-case examples and a Pi model-router build.
ProjectproductionPythonFree 12-notebook course building 13 agent tools with Jev as the fast decider and an LLM as the slow writer, for under $0.20.
ProjectproductionPythonBeginner manual, video script and three small examples for building an agent harness with Jev via langchain-typesafe.
ProjectprototypePythonVercel's decision guide on when to use a System One model like Jev versus a frontier model like GPT-6 Astra.
ArticlediscussionOpen-source AI gateway for 100+ LLMs that can use Jev to classify requests for complexity routing and to compact tool results.
ProxyproductionPython