Corv: an open-source AI-native SSH execution layer for infra agents

Dude01_ · reddit · 2026-09-17

A developer released Corv (v1.1.1), an open-source "AI-native SSH execution layer"—think MCP but purpose-built for infrastructure work. The author notes AI infra tooling gets little attention vs. developer tools, partly due to liability: you can't promise an agent won't nuke a system when deleting seems more efficient than fixing.

Key features: structured output instead of scraped terminal text; single connection, agentless, no tmux; safe retries via idempotency keys so a lost response never re-runs a mutating command; recoverable long jobs; credentials kept private via profiles; full audit logging of model-issued commands. It ships with skills for Claude and Codex and a staging environment for safe testing. The author claims that with top models like GPT6-Astra, an AI engineer that genuinely handles infra work is now feasible.

Related event: Corv: An Open-Source AI-Native SSH Execution Layer for Infrastructure(2 posts)→

Original post →

More from coding & agent

coding & agent channel →