CPU-only LLM inference engine in pure Rust — 4-bit quantized models, hand-written AVX2 kernels, speculative decoding, and a local HTTP/MCP server. No GPU, no Python runtime.
Local code-intelligence engine for AI agents — indexes a repository into a PostgreSQL code graph and serves token-budgeted context (symbols, effects, contracts, blast radius, similar code) over MCP and HTTP.
ClassEve Credential Airlock: self-hosted, OS-sealed credential firewall for AI agents. Agents see dummy keys; a deny-by-default local proxy injects real credentials only for allow-listed hosts and audits every request.
Lven downloads by ClassEve: Lven Instant (on-device voice-to-text) and Lven Cloud for Windows, Linux and Android, plus the speech engine the apps fetch. Versions, sizes and SHA-256 at classeve.com/releases/lven
Live speech translation for Android on your own Gemini or OpenAI key. No account, no subscription, and no backend — audio streams straight from your phone to the provider.
Outside witness: probes ClassEve's public surfaces every five minutes from GitHub's infrastructure and archives the result.
To associate your repository with the classeve topic, visit your repo's landing page and select "manage topics."