Agent session portability: start a task on one agent, finish with another

If your team is like ours, your engineers run several agents across a mix of enterprise and personal licenses. Ours cycles through Codex enterprise, Codex individual, Claude team, and Claude personal as we burn through token limits. We can't be the only ones who have hit a limit halfway through a piece of work and then had to rebuild the whole context in a second agent just to keep going.

Wouldn't it be better to port that context across and pick up where you left off?

As of today, Speakeasy Control Plane customers can. Session portability moves your agent session history from one agent to another, whatever the platform or license.

Context should be session independent

The vendors solved half of this already. Claude Code will --continue its own session and Codex will --resume, but both stop at the edge of their own platform and license. When a Claude Code thread hits a rate limit, or the next edit is visual and you'd rather make it in Cursor, you retype the investigation by hand into the second agent you already pay for.

The Speakeasy Control Plane captures full agent sessions across a company's managed devices. This release lets that captured context follow the work, from one agent to the next, on whatever tool the engineer wants to finish in.

How it works

Session portability runs on the Speakeasy Device Agent, and there are two ways to move context. The diagram below shows what happens between the two agents when you move one.

Continue from the menu bar

Open the Device Agent menu bar and choose Continue session. Recent sessions appear as two-line rows: what you were doing, which tool it ran in, how long ago, and in which repo. Pick one, then pick a target from the tools installed on the machine. The target opens in that session's project directory with a handoff that carries the task, the recent turns, the files you touched, the last assistant message, and an instruction to continue.

The CLI runs the same flow:

speakeasy sessions list
speakeasy sessions move <id> --to cursor

How the handoff arrives depends on the target. Claude Code and Codex start a new session with it as the first prompt. Cursor has no import API, so the workspace opens with the chat prefilled from SESSION_HANDOFF.md. If a launch fails, the files are already written and the exact command lands on your clipboard.

The session never leaves the machine

A move never touches the network. Device Agent reads the transcripts the tools already keep on disk, assembles the handoff locally, and uploads no session content. Each move writes a content-free audit event: who moved which session, into which tool, from which device. Your security team sees that the move happened without seeing the conversation.

How to get started

Confirm Device Agent is enrolled on the machine and Session Portability is enabled for your org, then use the menu bar flow above. Device Agent is in public beta, and this release runs on the macOS Device Agent fleet.

Rolling Device Agent across a fleet that already lives in mixed tools? Book time with our team.