Jev Ultrafast
A browser loop where Jev selects an operation and page element, with a text model handling field input.
A focused collection of projects and source materials.
Public projects, checked against their original sources.
A browser loop where Jev selects an operation and page element, with a text model handling field input.
An Android agent using Jev decisions and Mobilerun, with a live studio, CLI, and execution traces.
A macOS control loop that combines OCR and accessibility data with Jev action selection.
A staged Jev workflow for inspecting Git changes or source files, with a local review dashboard.
A Go CLI and MCP server that lets coding agents request typed Jev evaluations.
An official agent skill for designing TypeSafe workflows and finding the relevant API guidance.
A CLI and MCP browser tool pairing Jev decisions with agent-browser actions.
A Playwright browser harness splitting goal planning, typed action decisions, and text generation.
A Pi extension that adds Jev assessments before tool use and after file changes.
Hide page clutter with reusable, Jev-classified browser rules.
Delegate a bounded browser step to Jev through MCP.
Control macOS apps through accessibility targets chosen by Jev.
Bring Jev’s indexed action selection to the ego lite browser.
A compact Jev browser loop paired with a timed task benchmark.
Add Jev judgment labels to X posts and Hacker News comments.
One Playwright core exposed as a persistent CLI, SDK, and MCP server.
Explore native desktop control with Jev and an optional planning model.
Search and follow web evidence with Jev-ranked pages and links.
Inspect Jev browser decisions in a local visual control panel.
Let Jev drive browser steps and escalate uncertainty to the calling agent.
Run persistent Android tasks with Jev selecting valid device actions.
Add page checks and action judgment to an existing Playwright MCP setup.
Explore browser flows with Jev decisions and replayable test evidence.
Reorder Google results using Jev relevance and promotional-content judgments.
Optionally blur text flagged as advertising or AI-like prose.
Run browser goals through named actions and Jev selections.
Use Jev to supervise evidence from a coding agent’s work.
Evaluate Pi edits and actions against project rules with Jev.
Route Codex turns through Jev-selected model and reasoning settings.
Reduce large Claude Code tool results using Jev relevance judgments.
Expose typed Jev judgments as composable commands for coding agents.
Inspect a pull request’s risk, review depth, and test gaps.
Delegate file ranking, change checks, and triage to bounded Jev workflows.
Use a single model route with Jev selecting from a provider catalog.
Pair a Pi coding model with Jev judgments and explicit action policies.
Route Claude Code requests while accounting for prompt-cache tradeoffs.
Inspect MCP calls, results, and tool descriptions with Jev.
Review proposed database migrations with typed Jev judgments.
Add Jev option scores and optional skill routing to Claude Code.
Evaluate agent tool calls for risk, user intent, and untrusted influence.
Add an advisory Jev reviewer to an OMP coding session.
Run an issue-to-PR loop with tests, an independent reviewer, and Jev checks.
Combine repository policy with semantic change-risk signals.
Check a coding agent’s stopping point against plain-language rules.
Review whether a Pi session openly left required work unfinished.
Add semantic checks and traceable review signals to a single Pi agent.
Combine Jev safety signals and model selection in an LLM gateway.
Check source files against natural-language coding conventions.
Suggest a relevant installed capability for the current agent prompt.
Route an OpenAI-compatible model ID through a Jev decision.
Use Jev as a typed decision layer within Pi.
Give a Hermes agent a typed Jev tiebreaker.
Return structured code-quality judgments to a coding agent over MCP.
Try another keyword or clear your filters.