continues – Resume AI Coding Sessions Across Tools
Pick up an AI coding session from Claude, Gemini, Codex, or 13 other tools and hand it off seamlessly — no context lost.
TL;DR
TL;DR:
continuesscans your active AI coding sessions across 16 tools, picks one, and hands the full context off to another tool — Claude to Gemini, Codex to Cursor, any combination.
Source and Accuracy Notes
All information verified from the actual project resources listed below.
- Project page: github.com/yigitkonur/cli-continues
- npm package: npmjs.com/package/continues
- HN launch thread: news.ycombinator.com/item?id=47075089
- License: MIT (verified via README badge)
- Source last checked: 2026-07-23
What Is continues?
When you hit a rate limit mid-debugging session in Claude Code, you normally have two bad options: wait hours, or start fresh in another tool and lose all context. continues solves the handoff problem.
It discovers active sessions across 16 AI coding tools, reads each tool’s native session format (JSONL, JSON, SQLite, YAML — all different), extracts the conversation history and file changes, and generates a structured handoff document that the receiving agent can use immediately.
Supported tools: Claude Code, Codex, GitHub Copilot CLI, Gemini CLI, Cursor, Amp, Cline, Roo Code, Kilo Code, Kiro, Crush, OpenCode, Factory Droid, Antigravity, Kimi CLI, Qwen Code.
That’s 240 possible cross-tool paths.
Setup
No install needed — just run it with npx:
npx continues
Or install globally for faster access:
npm install -g continues # gives you `continues` and `cont` commands
How It Works
The four-step pipeline:
- Discovery — scans session directories for all 16 supported tools
- Parsing — reads each tool’s native format (they all store sessions differently)
- Extraction — pulls recent messages, file changes, tool activity, and AI reasoning
- Handoff — generates a structured context doc and injects it into the target tool
Usage Patterns
Interactive picker
npx continues
Finds all sessions across your tools, shows them in a TUI, and asks where to continue. Sessions from the current project directory are prioritized.
Quick resume within the same tool
continues claude # resume latest Claude session
continues codex 3 # resume 3rd most recent Codex session
continues amp # resume latest Amp session
Cross-tool handoff
# Hit Claude rate limit — hand off to Gemini:
continues resume abc123 --in gemini
# Pass flags through to the destination tool:
continues resume abc123 --in codex --yolo --search --add-dir /tmp
# Print the handoff prompt without launching:
continues resume abc123 --in codex --debug-prompt
--debug-prompt writes the handoff file then prints the exact prompt that would be passed to the target agent and exits without launching it — useful for inspecting what gets handed off.
Flag mapping
continues maps common flags (model, sandbox, auto-approve, extra dirs) to the target tool’s equivalent. Unknown flags pass through as-is.
FAQ
Q: Does it work with the web-based AI coding tools (Claude.ai, ChatGPT)?
A: No — continues works with CLI/agent tools that store session data locally (Claude Code, Codex, Gemini CLI, etc.). Web-based chat sessions are not accessible on the filesystem.
Q: How does it extract context from such different formats?
A: Each tool stores sessions differently. continues has a parser for each of the 16 tools — JSONL for some, SQLite for others, YAML or JSON for others. The README notes this is the hardest part of the project.
Q: Is session data sent anywhere? A: No. Everything stays local — it reads from your filesystem and generates a local handoff document.
Q: What license is it under? A: MIT, per the README badge.
Conclusion
If you juggle multiple AI coding tools — or find yourself constantly switching because of rate limits — continues eliminates the worst part of that workflow: starting over. Run npx continues and pick up exactly where you left off, in any tool.
Related Posts
ai-setup
Recall – Persistent Memory for Claude Code via MCP Hooks
Recall gives Claude Code a permanent memory store that survives session restarts and context compaction. Four hooks capture and restore context automatically — with cloud SaaS or self-hosted options.
2/28/2026
dev-tools
Automotive Skills Suite for AI Engineering
Evaluate Automotive Skills Suite for APQP, ASPICE, HARA, safety-plan, and DIA workflows with setup notes, governance risks, and SME review guidance.
5/28/2026
dev-tools
awesome-agentic-ai-zh Roadmap Guide
Explore awesome-agentic-ai-zh as a Chinese agentic AI learning roadmap, with setup notes, track selection, study workflow, and evaluation guidance.
5/28/2026