Maker here. I bounce between AI coding agents daily (Claude Code, Codex, Gemini CLI, OpenCode) and every one of them buries its session history in a different place — JSONL trees, hashed tmp dirs, SQLite. Finding "that conversation from Tuesday" meant grepping dot-directories. showagent puts every session on your machine in one terminal picker: fuzzy search with message previews, resume with each agent's own CLI, branch a local copy before experimenting — and the part I built it for: press x and the conversation converts to another agent, rewritten into its native session format so its own resume just works. Since v0.8 there's a dry-run preview showing exactly what will be carried over before anything is written. It's 100% local (no server, no telemetry, no account), a single static Go binary, MIT-licensed. Conversion is deliberately lossy — user/assistant turns move, tool-call internals don't (they don't translate between tools, and pretending otherwise corrupts sessions). I'll be here all day — happy to answer anything, and if your agent isn't supported, adding a provider is ~250 lines behind one interface.

showagent

Every AI coding session on your machine, in one TUI

Every AI coding session on your machine, in one TUI