A Handoff From Claude Code to a Fresh Agent

Claude Code writes a note to mountable.io. A fresh Claude Agent SDK reviewer reads the same note from its own runtime.

Claude Code finishes a task on your laptop and leaves a handoff note. A fresh reviewer in another runtime needs that note. Copying the laptop's files into each new job adds another step to every handoff.

mountable.io is a persistent filesystem for AI agents. The CLI and Python SDK reach the same files without a mount. Our Claude setup guide walks through a writer and a fresh reviewer.

Let Claude Code leave the note

Install the mountable skill in your Claude Code project. Then ask it to write the note to an existing filesystem:

Use the mountable skill to write "Parser review complete. Check empty
input next." plus a newline to /handoff.txt on filesystem FS_ID.
Use mountable files write with --if-none-match and stdin.
Read it back with mountable files read.

Use your filesystem ID in place of FS_ID. The command refuses to replace an existing note, so each task gets a new path.

Give the fresh reviewer just that note

The runnable Agent SDK example lets the reviewer read the selected note without offering a write tool. Your backend picks the filesystem and path before starting the agent. Run the reviewer from its own runtime:

uv run python review.py --filesystem "$FS_ID" --org "$ORG_ID" --note /handoff.txt

The reviewer also reports stored-byte usage, so the handoff's result includes how much the filesystem stores. It gets that figure through mountable.io's MCP server. The note itself comes through a read-only Python SDK handle.

We maintain this community example for mountable.io. It uses your existing mountable.io account and your own Anthropic credentials, with each service billing its usage.

Run the handoff with a short note from your next coding task.