Hi Product Hunt! 👋 A coding session can end with useful decisions, new constraints, and unfinished work. The next session needs that context. Runtive turns finished Claude Code and Codex runs into a record you can review and carry forward. Here’s the workflow: • Run `runtive init` once to connect your agent’s instruction files to `.runtive/context.md`. • After a task, run `runtive finish` to capture the session and git diff. Supply test, lint, or type-check output when you want it included. • Review the Decision Brief, then carry the extracted decisions, current state, open questions, and next steps into your next session. Context entries trace back to the runs they came from. The Decision Brief also marks dimensions it cannot verify. If extraction fails, the captured run remains available to retry. Runtive Workspace is $19/month with a 14-day free trial. Extraction uses your own model credentials; model usage is separate from the subscription. The Anthropic API key route is the production path, while local Claude Code extraction is currently beta. What context do you most often have to repeat when starting a new coding session? I’d love your feedback.

Runtive

Keep Claude Code and Codex in context across sessions

Keep Claude Code and Codex in context across sessions