Jev ecosystem projects — page 143

Interviewer-side live signal layer for mock interviews, built on TypeSafe's Jev. Typed judgments, a suppression policy, and an honest usefulness gate.

TypeSafe jev-powered LLM model router: classify → policy → tiered pools → OpenAI-compatible passthrough proxy (streaming, tools, full agent loops)

Why an agent stopped and how risky its prompt is, as a badge on its pane: TypeSafe's Jev reads the screen modisa's rules can't.

moonbit CI gate with jev

AI code review against your repository's rules, powered by TypeSafe AI's Jev.

Identify a pressing environmental problem affecting your community or the world and develop a creative solution to address it.

Jev-graded tool approvals for oh-my-pi: suppress the approval prompts you did not need, keep the ones that matter.

omp extension: a read-tier `decision` tool backed by TypeSafe System One (Jev) — typed choice/noul/score questions over any state, answered with probabilities and token cost

OMP browser tool that drives any page from one goal using TypeSafe Jev (System One) decisions instead of the main model.

Reproducing the real $1 Chevy Tahoe chatbot exploit and testing 6 AI defenses (incl. TypeSafe's real Jev API) against 36 prompt-injection attacks.

High-speed System One Jev AI decision gateway and MCP server powered by OpenRouter

Open System One model for Apple Silicon: typed decisions with calibrated probabilities from one forward pass.

A command game where the enemy is what people think you meant. Orders in your own words, measured by TypeSafe Jev; officers with doctrine decide.

Open teaching pack for TypeSafe Jev in Grok Bot — three lanes, setup, examples. Gold Hat. No secrets.

Chrome extension: Jev (TypeSafe) tags Outlook Web emails as worth reviewing and categorizes them (PO Variance, Patch Notes, Items, Cost Containment)

Verbatim Jev context compaction for Pi Coding Agent

Single-file pi extension for batched TypeSafe Jev judgments

Browser and macOS desktop agent for pi: Jev (TypeSafe System One) chooses each action from a structured observation in a bounded, surface-agnostic loop. Isolated Playwright tools, an allow-listed accessibility-tree tool, deterministic selectors, four-tier benchmarks.

Pi extension that replaces the built-in LLM compaction summary with TypeSafe Jev's per-tool-call scoring (lossless pruning instead of summarization).

Session-pinned model and thinking selection for Pi using TypeSafe Jev.

Benchmark for Pi skill routing: BM25 vs TypeSafe Jev across roster sizes 50–500