Skip to main content
Interactive. Lets you pick which coding agents ResumeContext reads history from, and where their data lives on disk if it is not in the default location.

​
Without asking

Two flags cover the same ground with no prompt, for a script or a coding agent acting for you:
--list needs no project — it reports every supported agent, the directory its history normally lives in, whether that directory holds history on this machine, and, if run inside a project, what that project currently syncs from. It is what lets someone else put the choice to you before setting it. --set replaces the choice outright, taking name or name=/path/to/history entries. A directory that does not hold that agent’s history is refused, the same as when typed into the prompt.

​
Without asking

Two flags cover the same ground with no prompt, for a script or a coding agent acting for you:
--list needs no project. It reports every supported agent, the directory its history normally lives in, whether that directory holds history on this machine, and — inside a project — what that project currently syncs from. That is enough to put the choice to someone before setting it. --set replaces the choice outright, taking name or name=/path/to/history entries. Repeat an agent to give it several directories, exactly as the prompt allows:
A directory that does not hold that agent’s history is refused, the same as when typed into the prompt.

​
Supported agents

Nothing is installed into the agent itself. Each of these already writes its session history to disk, and the daemon reads that, so there is no plugin and nothing to change about how you prompt.
Those are the default locations, and they are a starting point rather than an assumption. If yours keeps its history somewhere else, agents will take the path and check it looks right before accepting it.
Support is per agent, and it amounts to teaching the CLI where one keeps its history, so this list grows. If the agent you use is missing, that is the thing to ask for.

​
When to run it

  • You start using a new agent and want its sessions archived.
  • You want to stop archiving one.
  • Your agent stores its history somewhere non-standard.
The choice is per project, because a project’s archive should contain the work done on that project, whichever tool did it. Two people on the same project can be using entirely different agents and still read each other’s sessions.
agents is the one project command that does not start the daemon. It only edits configuration that a running daemon re-reads on its next tick anyway, so starting one here would silently undo a daemon stop you meant.