What does Jev do inside JIT-JEV Context OS?

Jev acts as the System 1 decision gate: it shadow-evaluates facts in the compiled context capsule, proposes evictions before context bloats and prefetches relevant memory before agent turns. A fail-open circuit breaker (invariant I6) keeps the harness running when Jev is unavailable.

What benchmark results does the repository publish?

The fixture-table run reports 31.3% fewer turns (6.7 to 4.6) and 52.6% fewer blind discovery operations across 10 SWE-bench-style tasks. A separate 45-module production snapshot claims 2.44x faster delivery and a >99% prompt-token reduction; the README explicitly separates those runs.

What runtime and target environment does JIT-JEV Context OS need?

The L0 memory layer runs on SQLite WAL with a claimed sub-3ms budget between agent turns. It is built for autonomous coding agents with active tool execution, such as file read/write, patch, terminal and AST parsing, not for generic chat-only platforms.

Can JIT-JEV Context OS run fully offline with a local model?

Yes. In its controlled multi-file repair test a local Qwen 3.8 9B on a Mac Mini M2 Pro with 32k context solved 4/4 verified tests in 4 turns, fully offline at $0.00. A Google Gemini run without JIT failed all four on the same task.

How large is the context capsule it compiles?

The JIT Context Compiler builds a prefix-aligned Context Capsule of under 1,500 tokens for each execution turn instead of stuffing 50,000+ tokens of raw logs and files. In the 45-module production snapshot the active prompt footprint was 482 tokens.

Does the Jev gate need an API key, and what happens if Jev is down?

The live evidence was produced with typesafe/jev-1.13-20260917, so the System 1 gate uses the hosted Jev model and needs a TypeSafe key. Invariant I6 is a fail-open circuit breaker, so the harness keeps running rather than blocking when Jev is unavailable.

Where is the project published, and what invariants does it document?

It is MIT-licensed and published on Zenodo under DOI 10.5281/zenodo.22649542, with epistemic invariants I1 through I10. Its benchmark files separate live Jev agent-loop runs from heuristic runs, and the README notes the SWE-bench-style run uses fixture trees rather than full checkouts.

CommunityJev Integrations & Adapters

JIT-JEV Context OS

Epistemic runtime for autonomous coding agents that uses Jev as a sub-100ms System 1 gate to shadow-evaluate capsule facts, propose evictions before bloat and prefetch memory between turns. Its fixture run reports 31.3% fewer turns and 52.6% fewer blind discovery operations, with a fail-open circuit breaker.

See highlights

Open github.com

Stars: 9
Language: Python
License: MIT
Published by: Community
Last verified: Oct 2026

wojciechwiesner/jit-context

User repository on GitHub

View on GitHub

Architectural Moat: JIT-JEV Context OS — Epistemic runtime & JEV System 1 context gate for AI agents (L0 SQLite WAL <3ms, Epistemic Invariants I1–I10, CERN Zenodo DOI: 10.5281/zenodo.22649542)

Forks: 1
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.Homepage

About this project

The Epistemic Operating System for Autonomous Coding Agents. Eliminating the Haystack Tax: 61% fewer agent turns, 2.44x faster delivery, and >99% prompt token reduction on production codebases. JIT-JEV Context OS is an epistemic context runtime and fast decision engine built specifically for autonomous coding agents with active tool access. It bridges two complementary layers:

Quoted from the project's README. Read the full README.

Tagscommunity agents context-engineering swe-bench routing

Works forJev for Scoring Jev for Decision gating Jev for Routing

Used inJev for Software & DevTools

Languages: Python
Platforms: SQLite
Capabilities: Open sourceRequires API key

Highlights

L0 memory layer runs on SQLite WAL with a claimed sub-3ms budget between agent turns.
Production snapshot on a 45-module codebase: 66 vs 171 agent turns and 482 vs 50,000+ prompt tokens.
SWE-bench-style fixture run: 4.6 vs 6.7 turns per task and 18 vs 38 blind discovery operations.
Publishes on Zenodo (DOI 10.5281/zenodo.22649542) and documents epistemic invariants I1-I10.

Watch out

MIT licensed. It targets autonomous coding agents with active tool execution and is not a generic chatbot wrapper, and several production numbers come from the author's own harness runs. The repo itself notes its SWE-bench-style run uses small fixture trees, not full repository checkouts.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

Facebook: Intake Router keeps diagnosis out of the answer space at the front desk.

FacebookClinicnotes

YouTube: Jev explained for routing and judging, with a comparison against LLMs.

Hacker News: Vercel on using Jev for defined decisions and GPT-6 Astra when the job also has to write or use tools.

flashbrew4 pointsTypeSafe AI Jev vs. GPT-6 AstraRead the thread on Hacker News

Hacker News: ASTGL replay chose local Laya routing over hosted Jev on one Mac Studio setup.

Jmeg8r15 pointsDoes an open-weight decision model beat a hosted one? Jev vs. LayaRead the thread on Hacker News

Related terms

Glossary definitions related to this entry.

Intent routing

More Jev resources like this

AutoGPT

187.5k

Autonomous agent platform with 187,504 stars for building and running agents from plain English or a visual builder. Its platform features first-class TypeSafe Jev decision blocks for typed routing, filtering, scoring and confidence-gated next-action dispatching.

Jev Integrations & Adapters#community#autonomous-agents#platform

Community

Significant-Gravitas/AutoGPTPython

AgentScope

33k

Alibaba's multi-agent platform, which implements native TypeSafe Jev classification through JevClassifierModel for binary, choice and score routing across agent pipelines. Python 3.11+, Apache-2.0, with a FastAPI agent service and web UI.

Jev Integrations & Adapters#community#multi-agent#framework

Community

agentscope-ai/agentscopePython

Pydantic AI

20.5k

Python agent framework whose TypeSafeModel class runs Jev as a decision model inside Pydantic AI agents: Pydantic output-type fields become typed Jev questions answered together in one request, each answer carrying its own confidence, and tools can be offered as Choice options.

Jev Integrations & Adapters#community#agents#framework

Community

pydantic/pydantic-aiPython

GoEventBus

High-performance Go event bus with a bounded MPMC ring buffer, fan-out and ordered handlers, plus an optional routing layer that evaluates deterministic rules first, reuses cached decisions, and asks Jev a typed Choice only for ambiguous cache misses.

Jev Integrations & Adapters#community#go#event-bus

Community

Protocol-Lattice/GoEventBusGo

jev-oncall

On-call tooling where Jev answers four typed questions per alert - actionable, severity, owning team and duplicate-of - and plain code routes, pages or drops. Includes a root-cause benchmark where Jev and code investigate recorded incidents with no general-purpose LLM in the loop, enforcing budgets and verification.

Jev Integrations & Adapters#community#alerting#incident-response

Community