i kept hitting rate limits in Claude Code mid-debugging, then hopping to Gemini or Codex. the annoying part wasn't switching tools (copy-pasting terminal output doesn't bring tool-use context with it) — it was losing the full conversation and spending 10 minutes re-explaining what i was doing.

so i built *continues*. it finds your existing AI coding sessions across five tools (Claude Code, GitHub Copilot, Gemini CLI, OpenAI Codex, OpenCode), lets you pick one, and generates a structured handoff so you can continue in another tool without starting from zero.

    npx continues                          # interactive TUI: pick a session, pick a target
    continues scan                         # see what it finds (read-only)
    continues claude                       # jump into your latest Claude Code session
    continues resume abc123 --in gemini    # hand off a specific session to Gemini

the flow:

* scans the current directory first (so you see what's relevant), then shows everything

* you pick a session + a target tool

* it generates a handoff doc: recent conversation, cwd, files modified (best-effort), pending tasks

* launches the target tool with everything injected inline — no extra "go read this file" step

the main engineering pain was that every tool stores sessions differently (and none of the schemas are standardized):

* Claude Code → JSONL in `~/.claude/projects/` (includes `isCompactSummary` messages you don't want in the handoff)

* GitHub Copilot → YAML metadata + events JSONL in `~/.copilot/session-state/` (no user messages logged; mostly assistant/tool events)

* Gemini CLI → JSON in `~/.gemini/tmp//chats/` (content shape varies; tool-call-only messages exist)

OpenAI Codex → JSONL rollouts in `~/.codex/sessions/` (dual `event_msg` / `response_item` format)

* OpenCode → SQLite via Drizzle ORM in `~/.local/share/opencode/` (recently migrated from JSON, which broke things once already)

each tool also wants prompts differently (positional args vs flags like `-i` / `--prompt`), and Gemini CLI sandboxes file access to the project dir. getting all the cross-tool paths (5×4 = 20 conversions) working was most of the work.

what it's not:

* not "true migration" — it's context injection. you get recent messages + metadata, not a full replay

* rate limit detection is manual for now — you run it when you know you're blocked, no auto-detect yet

* session formats are mostly undocumented and can change anytime (this is the biggest maintenance risk)

* local file parsing only, no API calls — your data stays on your machine

curious if anyone else actually juggles multiple AI coding CLIs, or if most people just commit to one and wait out rate limits. would love to hear how you handle tool-switching + context today + feedbacks on context quality migrations and feedbacks are welcomed.

Gemini 3.1 Preview now live in AI Studio

Gemini 3.1 Pro

Could Write­Process­Memory be made faster by avoiding the intermediate buffer?

Baseline Drift

Unlock Unblocked Music YouTube: Easy Guide

Where will the amounts of water come from?

Show HN: Crit – Visual QA for iOS apps and AI coding agents

Federating Databases with Data API Builder Chaining

Show HN: How far can you push file conversion into the browser?

Open source protocol that improves AI code quality in any IDE

Epitome – Open-source shared memory layer for AI agents

XC-Basic Shoot 'Em Up: Programming the Pet Part 2

Show HN: Bravehuman – AI training for high-stakes conversations

Fire calculator that takes Bitcoin holdings into account

Parents say teens are addicted to social media. Now, a jury will decide

Biota · a Personal Catalog of Life

Show HN: Micasa – track your house from the terminal

Powering the next generation of agents with Google Cloud databases

Members-only Philly cop bar has been linked to two DUIs and a third crash

A Practical Security Audit for Builders

Show HN: Voice AI that answers your restaurant's phone and posts orders to POS.

Ask HN: My post does not show up in /newest again

Show HN:`npx continues` – resume same session Claude, Gemini, Codex when limited

Why Not Objective-C

The End of Lazy SaaS: Same Game, Harder Rules

Super stable laser on the moon could guide future lunar missions

Markpub.at Markdown Lexicon

Problems in E-Commerce Ripe for SaaS Builders

Freedom.gov: US State Department Plans VPN Portal for Europe

Amid Epstein fallout, Bill Gates becomes point of controversy at India AI summit

Show HN: Decks For Good – get fundraising advice when you give to charity