Why use a separate CLI instead of calling Jev directly?

It isolates small, bounded judgments behind a standard-library-only interface with explicit consent gates, input limits and a documented output contract, so the calling agent keeps reasoning, verification and action authority while Jev answers closed-set, yes/no, rating or extraction questions.

What happens if Jev is unreachable?

Missing key or consent, 401/402/429 responses, connection failures, malformed responses, storage failures and deadline expiry all return a fallback with exit 0. Invalid local input returns a structured error with exit 64, so scripts can distinguish bad input from advisory degradation.

What are the size limits on a request?

256 KiB of input, 96 KiB of encoded request, 28 KiB per state/question pair and a 32 KiB response. Batch questions share those limits and one deadline, and an oversized batch is not split into extra network requests, so a batch that will not fit is rejected rather than quietly split.

How do the local thresholds decide between an answer and none?

choose returns an answer only when both confidence and winner probability reach 0.85. check returns true at or above 0.85, false at or below 0.15, and unknown in between. score returns the ordinal value only at 0.85 confidence or above. These are local operating policy, not calibrated accuracy guarantees for your data.

What does batch send, and what comes back?

One state and 1 to 32 independent Choice, Noul or Score questions in a single request. Answers are keyed by question id, one question cannot read another's answer, and the caller picks the branch to apply afterwards. An expect value is local policy stripped before the provider request so it cannot suggest the answer.

What does purify return, and where does the text go?

Selected original spans from the source with local evidence and no generated summary. Raw text is saved privately before evaluation, successful results expose receipt and evidence_path pointers, and source hashes establish identity rather than truth, so consult the retained source when passage selection may have missed context.

How do I install it and what has to be configured?

Python 3.10 or newer on POSIX, with no third-party runtime dependencies, GPL-3.0-or-later. The key is read from the process environment as TYPESAFE_API_KEY and never from a credential file, and remote calls also need the exact consent value HARNESS_JEV_ALLOW_REMOTE=1, since the literal true is refused. JEV_MODEL and HARNESS_JEV_EVIDENCE_DIR are optional.

CommunityJev Tools & CLIs

Ask Jev — logicrw

Zero-dependency Python CLI that routes small semantic judgments to Jev for agents and CLI pipelines: choose (Choice), check (Noul), score (ordered ratings), batch (1-32 questions in one request) and purify (verbatim passage extraction). GPL-3.0 licensed.

See highlights

Open github.com

Stars: 2
Language: Python
License: GPL-3.0
Published by: Community
Last verified: Oct 2026

logicrw/ask-jev

User repository on GitHub

View on GitHub

Ultra-fast, fail-open advisory decisions and verbatim extractive reading view for AI coding agents and CLI pipelines

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

Bounded advisory decisions and verbatim reading views for AI agents. English · 简体中文 · Output contract · Code triage patterns · Awesome Jev. ask-jev delegates small semantic judgments to TypeSafe's System One model through a Python standard-library CLI. It supports closed-set selection, evidence checks, ordered ratings, mixed questions over one state, and verbatim passage selection. The calling agent retains reasoning, verification and action authority.

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

Tagscommunity cli decision-routing fail-open

Works forJev for Classification Jev for Extraction Jev for Decision gating

Used inJev for Software & DevTools

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

Highlights

Five commands: choose, check, score, batch and purify, returning JSON under the schema identifier ask-jev.v1.
Remote calls require both TYPESAFE_API_KEY and the exact consent value HARNESS_JEV_ALLOW_REMOTE=1; the literal true is not accepted.
An advisory 280ms budget covers evidence preparation, one HTTP request and receipt persistence; there are no retries, daemons or listening ports.
Local thresholds: choose needs confidence and winner probability >= 0.85; check returns true >= 0.85, false <= 0.15, otherwise unknown.
Zero third-party runtime dependencies on Python 3.10+ POSIX; provider and storage failures fail open with exit 0, while invalid input exits 64.

Watch out

GPL-3.0-or-later. It is fail-open by design: missing consent, auth errors or deadline expiry silently return fallbacks, so exit 0 alone does not prove Jev ran.

Questions about this entry

Short answers drawn from the linked project or page.

Related terms

Glossary definitions related to this entry.

Fail open

Other listings of this project

Forks and copies of this project in the directory. This entry is the canonical one.

ask-jev

A PowerShell hook for Codex that intercepts your prompt turn and asks Jev to audit the conversation recorded so far, answering with typed probabilities rather than prose.

Jev Agents & MCP#community#jevbest#catalog-import

Community

omni-/ask-jevPowerShell

More Jev resources like this

grev

grep, sort, cut and uniq that match by meaning: each line whose Jev Noul clears the threshold survives grev, sibling filters route by Choice and rank by Score, and the output is always your own input, never generated text.

Jev Tools & CLIs#community#cli#coreutils

Community

aurorainfra/grevGo

Edward

Supervisor for unattended coding agents: it tracks cross-turn state and asks one batched Jev Choice whether a trajectory should continue, pause or escalate, while deterministic code owns dangerous-command blocking, budget caps and Ed25519-signed receipts.

Jev Tools & CLIs#community#guardrails#safety

Community

VeridicalTech/EdwardPython

GUI JEV Harness

Python harness for GUI grounding: it divides a screenshot into grid tiles, has a vision model describe each tile neutrally, then asks typesafe-ai/jev to choose among the descriptions. Local confidence and margin gates decide whether to descend or refuse, and it emits only a point and bounding box.

Jev Tools & CLIs#community#gui-grounding#computer-use

Community

ZihuaEvan/GUI_JEVPython

jev-research-pipeline

A CLI that checks new papers, repositories and feeds each morning against your open research questions, asking Jev Noul gates and Score dimensions through Pydantic AI's typesafe model; an LLM writes only the prose into Obsidian notes.

Jev Tools & CLIs#community#research#cli

Community

shimo4228/jev-research-pipelinePython

jeff

Go CLI for Jev with noul, choice, score, eval, rank, and models commands: rank sends one Score question per item per weighted YAML dimension in a single request and sums weight times score in code, while threshold flags turn a failed gate into exit code 10 for scripts and CI.

Jev Tools & CLIs#community#cli#go

Community

saembit/jeff-cliGo

TradingAgents

110.5k

Multi-agent trading research with optional Jev screening of StockTwits and Reddit posts for company relevance before sentiment analysis runs.

Jev Tools & CLIs#community#python#trading

Community

TauricResearch/TradingAgentsPython

From the Jev guides

Original write-ups on what this entry is for.

Jev on the Gateways: Vercel, Workers AI, and OpenRouter

Oct 12, 2026

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

Exporting Decision Models: GGUF, ONNX, and Device Runtimes

Oct 10, 2026

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

Records at Rest: Documents, Inboxes, and Risk Queues

Sep 28, 2026

Jev labels a record that already exists. Code moves, flags, or holds it. How the document, mail, and transfer tools keep the action out of the model.

Jev Tools & CLIs#classification#email#security

Read article

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai