Jev Proxy

A router that asks a classifier, Jev by default, how much reasoning each step of a coding session needs, then adds that effort to the model request without breaking the prompt cache. It ships as an OpenCode plugin, a Pi extension and a local proxy for Codex CLI.

A self-hostable HTTP proxy that fronts TypeSafe Jev through one endpoint, rotating your API keys automatically and serving the /v1/systemone wire format. Keys live in Upstash Redis or an environment variable, and a web admin page manages them.

LiteLLM serves decision models on two routes that share one provider list: /v1/decisions for OpenAI-format bodies and /v1/systemone for System One bodies, with Jev available as the typesafe/jev-latest provider.

A lightweight, self-hosted LLM proxy that gives Hack Club's teen community free chat, embedding, and image models over OpenAI-compatible endpoints, with Hack Club auth, API keys, moderation, spending limits, and analytics.

A host-agnostic routing layer that sits between a coding harness and its models, choosing capacity per round from trajectory evidence like tool calls, failures and context pressure, running as a user-scoped daemon and local OpenAI-compatible proxy.

Route OpenAI Codex CLI turns through Jev: a local loopback Responses API proxy asks Jev to pick a Codex model and reasoning effort for each fresh turn, then forwards the request. If Jev is unavailable, Codex continues unchanged.

A governed execution layer and drop-in proxy between typed-decision models like Jev and callers such as Claude Code, Cursor/MCP and SDKs: raw probabilities become deterministic policy actions from ALLOW to DENY with a DuckDB audit trail.

Use Jev from Claude Code, Codex CLI, Claude Desktop or any MCP client, paid for by free Netlify credits and locked with a key only you have: a key-locked proxy forwards through Netlify AI Gateway and an MCP server exposes jev_evaluate and jev_models.

A single-file Python proxy that asks Jev which of four model tiers each new Codex turn needs — luna, terra, sol or astra — rewrites the model field, and retries flaky upstream errors.

A drop-in proxy that distills a repeated Jev classification call into a local model as it runs: it learns from Jev's own probability distributions, then answers most requests itself while uncertain ones and a 2% audit slice keep going to Jev.

Testable 48-requirement spec plus conformance suite, normalising proxy and fault-injecting mock for Jev-compatible System One API servers; against the eight most-starred open servers only two passed every MUST requirement.

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.