Pyromind-Dynamics/pi-jev-governor
Organization repository on GitHub
- Forks
- 0
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Pi JEV Governor
A decision governance layer for the pi coding agent: planning, model routing, permission control and outcome validation through JEV. Claims up to 45% lower model cost on MATH-500 with no accuracy loss and P50 decision latency of 0.6–0.7 s.
- Stars
- 1
- Language
- TypeScript
- License
- MIT
- Reading time
- 2 min
- Published by
- Community
- Author
- Pyromind-Dynamics
- Last verified
- Oct 2026
Used inJev for Software & DevTools
- Languages
- TypeScript
- Platforms
- CLI
- Capabilities
- Open sourceRequires API key
Highlights
- Governs the pi agent lifecycle: how a task starts, which model handles it, whether tool actions proceed, whether it can finish.
- Claims up to 45% lower model cost on MATH-500 with no accuracy loss; Jev overhead is 1.4–2.6% of cost.
- Reports decision latency P50 of 0.6–0.7 s through the JEV Decisions API; requires pi 0.87.1+ and TypeScript 5.x.
- Design rule: deterministic rules handle hard boundaries, JEV handles bounded decisions, humans handle low-confidence and high-risk cases.
- MIT licensed with a CI workflow in the repo; created 2026-10-13 and pushed 2026-10-14, per GitHub metadata.
Watch out
README and GitHub metadata reviewed 2026-10-17; the digest ends at the MATH-500 results image, so the benchmark methodology, install steps and configuration are not captured. The cost and latency figures are the project's own claims.
Questions about this entry
Short answers drawn from the linked project or page.
A decision governance layer for the pi coding agent, powered by JEV, covering planning, model routing, permission control and outcome validation across the agent lifecycle.
Four: how a task should start, which model should handle it, whether tool actions may proceed, and whether the task is actually ready to finish.
Up to 45% lower model cost on MATH-500 with no accuracy loss, Jev decision overhead of 1.4–2.6% of cost, and decision latency P50 of 0.6–0.7 s, per the README.
Its stated design: deterministic rules handle hard boundaries, JEV handles bounded decisions, and humans handle low-confidence and high-risk cases.
pi 0.87.1+ and TypeScript 5.x per the README badges, plus access to the JEV Decisions API; the repository is MIT licensed with CI configured.
They are the project's own README claims; the digest ends at the MATH-500 results image, so the methodology behind them is not visible for review.
Pyromind-Dynamics; the repository has 1 star, was created 2026-10-13 and last pushed 2026-10-14, per the digest metadata.
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
X: OpenRouter Decision Model Rankings
X: herdr + jev, an agent router
X: A group chat bot with decision room routing
More Jev resources like this
An oh-my-pi extension that puts consequential agent decisions to Jev (jev-latest) through the TypeSafe API, checking visible actions and completion boundaries. Its ordinary-session entry is deliberately disabled: an MVP under verification, not a release.
Jev Tools & CLIs#community#github#pi
Community
veschin/jevsticeTypeScript
veschin/jevsticeTypeScriptAn ultra-lean supervisor extension for the pi coding agent: zero-token regex guardrails stop destructive commands, a done-check tracks test verification, and a Jev semantic gate reviews diffs before disk writes.
Jev Tools & CLIs#community#catalog-import#logicrw
Community
bismawy/pi-jev-eyeTypeScript
bismawy/pi-jev-eyeTypeScriptMIT-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
A Claude Code mod that scores each prompt with TypeSafe Jev and sets Claude's effort level for that turn, mapping four narrow questions, reasoning depth, scope, stakes and ambiguity, onto an effort level without ever touching the model.
Jev Tools & CLIs#community#github#claude-code
A TypeScript/Bun launcher that routes coding-agent user turns through JEV, the TypeSafe classifier, with model and reasoning-effort selection: codex-jev wraps the Codex CLI, claude-jev wraps Claude Code, and a Pi extension adds a jev/auto virtual model.
Jev Tools & CLIs#community#github#cli
A Claude Code mod that routes each request to haiku, sonnet or opus with a reasoning effort and confidence score using local heuristics, and can optionally let Jev (TypeSafe System One) pick the tier instead. Adds a live status band above the prompt.
Jev Tools & CLIs#community#github#claude-code
Community
visualife/ClaudeRouterTypeScript
visualife/ClaudeRouterTypeScriptFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.