Hydrate for OpenAI Codex CLI · v1 launch
Your Codex CLI, with portable memory.
Hydrate is a measured, auditable, on-premise memory and efficiency layer for AI assistants. The same binary that ships for Claude Code extends to the OpenAI Codex CLI via a thin shim — facts captured on one tool surface in the next prompt on the other.
What works today
- UserPromptSubmit-style hook wired through the Codex CLI lifecycle.
- MCP server reachable as
hydrate-mcp; Codex CLI consumes it as a standard MCP client. - Shared canon and facts with any other Claude Code / Cursor / Cline / Gemini CLI session on the same machine.
What's rough
- Capture coverage is partial — some Codex CLI tool calls don't yet emit observations Hydrate can index.
- Compaction behaviour differs from Claude Code's; the PreCompact equivalent is not yet wired.
- Install surface is manual today; no
hydrate codex installsubcommand yet.
We're publishing alpha builds because the shared-memory story is the interesting part and we'd rather you try it than wait. Production-grade Codex support ships with Hydrate v1.
Ships with Hydrate v1 · currently alpha
In open beta — install live.
The Codex CLI integration is in alpha today and ships production-grade alongside Hydrate's public v1 release. Press, researchers and anyone writing about the launch can request early access.