User repository on GitHub
A pi package that gates subagent dispatches through a JEV System One decision model, plus general-purpose tools to query that model.
- Forks
- 0
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Tools & CLIs
Pi Subagent Jev
pi coding-agent package that gates subagent dispatches: the task text is evaluated in one typed Jev call against configurable rules (per-rule probability questions with below/above thresholds), violations block the dispatch with per-rule reasons, and errors fail open to allow.
- Stars
- 2
- Language
- TypeScript
- License
- MIT
- Published by
- Community
- Last verified
- Oct 2026
About this project
A pi package that gates subagent dispatches through a JEV System One decision model, plus general-purpose tools to query that model. Dispatch compliance gate — when the main agent dispatches a subagent, the task text is sent to the JEV System One endpoint for evaluation. If any configured rule fires, the dispatch is blocked: every violation message is returned to the main agent as an error result, and the subagent is never spawned.
Quoted from the project's README. Read the full README.
- Languages
- TypeScript
- Capabilities
- Open source
Highlights
- The dispatch gate hooks the subagent tool from pi-subagents: on a hit the subagent is never spawned and violation messages return to the main agent.
- Rules live in ~/.pi/agent/jev-comp/compliance-rules.json; the code ships no built-in rules, so without a config file no checks run at all.
- Every evaluation, allowed or blocked, appends one line to ~/.pi/agent/jev-comp/audit.jsonl; task excerpts are truncated to 200 characters.
- jev_ask evaluates a batch of noul/choice/score questions against a state text, with optional named-upstream selection and an ms latency field.
- v0.8.0 changed the rules schema, and optional runtime monitoring watches for bash stalls, tool loops and repeated failures without interrupting the child.
Watch out
MIT licensed, but it currently supports only pi together with the pi-subagents package, and v0.8.0 introduced a breaking rules-file schema change. With no rules file it performs no checks at all, by design (fail-open).
Questions about this entry
Short answers drawn from the linked project or page.
Every violation message is returned to the main agent as an error result and the subagent is never spawned. Each evaluation is also written to the audit log with agent, verdict, rule ids, model and latency, plus raw probabilities only when auditProbabilities is enabled.
Yes. A group can set mode: warn, or _global.mode sets a default, so a violating dispatch is not blocked -- the violation list is appended as a trailing notice and the main agent judges for itself. Invalid mode values silently fall back to block.
Run `pi install git:github.com/G0-0000/pi-subagent-jev`, then create ~/.config/jev-comp/env with permissions 600 holding JEV_AI_API_KEY and JEV_AI_BASE_URL (JEV_AI_MODEL is optional, default oc/jev-1.13-free). Run `/reload` or restart pi to pick it up.
It runs but checks nothing. Since v0.7.0 the code ships no built-in rules, so an install without compliance-rules.json performs no checks at all. Invalid, unknown or duplicate rule references are silently ignored, so the design is fail-open.
Every evaluation, allowed or blocked, is appended as one JSON object per line to ~/.pi/agent/jev-comp/audit.jsonl, with the agent, a 200-character task excerpt, model, rules, verdict and latency. API keys are never logged, and raw probabilities appear only when _global.auditProbabilities is enabled.
It queries the decision model directly: you supply state text and a map of typed questions (noul, choice or score) and it returns the answers. Both state and questions are optional and fall back to a built-in probe payload, and an optional upstream name sends one request straight to a named upstream.
Yes, opt-in runtime monitoring watches running subagents for suspected bash stalls, unproductive tool loops and repeated failures, and nudges the main agent with advisory alerts. It never interrupts, and it is off by default.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Agent Handoff Gate: agents verify what they hand off
X: Jev as an LLM guardrail inside Agentgateway
X: Vercel's fx safety reviewer, measured at 18× faster
Reddit: Early experiments wiring Jev into Pi as a safety layer and model router.
More Jev resources like this
27GitHub stars
A DeepSeek Harness plugin where Jev Choice and Noul checks gate every non-read-only tool call - deny confident high-risk calls, escalate ambiguous ones, auto-approve clearly granted reversible ones - and optionally score session messages so valuable context survives instead of being dropped oldest-first.
Jev Tools & CLIs#community#plugin#guardrails
13GitHub stars
Permission gate extension for the pi coding agent: deterministic allow/deny rules and protected paths settle clear cases first, then gray-zone tool calls get one structured classify() verdict; Jev is an optional native classifier via TypeSafe direct, OpenRouter, or other gateways, failing closed on error.
Jev Tools & CLIs#community#guardrails#permissions
Community
jesset/pi-verdictTypeScript
jesset/pi-verdictTypeScript8GitHub stars
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
VeridicalTech/EdwardPython0GitHub stars
OpenClaw plugin that checks every outgoing agent message against its destination channel tier, running local key-format and term rules and then five Jev Noul questions in one call through OpenClaw's decisionModel, which can point at hosted Jev or a local Kev.
Jev Tools & CLIs#community#openclaw#guardrails
0GitHub stars
OpenAI-compatible /v1/chat/completions proxy that answers an agent host's approval-guardian call with APPROVE, DENY or ESCALATE. Jev answers one Choice question with a calibrated distribution; missing keys, timeouts and malformed answers all fail closed to ESCALATE.
Jev Tools & CLIs#community#agent-safety#guardrails
Community
oppih/approval-judge-bridgePython
oppih/approval-judge-bridgePython39.8kGitHub stars
A teams-first multi-agent orchestration plugin for Claude Code with opt-in Jev hooks for sub-millisecond judgment points, decision caching and per-point egress controls. TypeScript, MIT, with a staged team pipeline and tmux CLI workers.
Jev Tools & CLIs#community#claude-code#orchestration
From the Jev guides
Original write-ups on what this entry is for.
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
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
From the Jev community
Posts from builders shipping with Jev right now.