When several hunk sessions exist, everything funnels through a session id. In practice you only care about this git checkout. Right now you either pick an id by hand, or pipe hunk session list --json through jq to match repoRoot. That’s awkward for humans and worse for automated agents.
Let hunk session subcommands (and ideally hunk session list) accept something like --repo path name and --worktree path name (defaulting to the current repo) so the CLI resolves the right session.
Same behavior people already get by filtering the session API; the CLI should make the common case one command.
When several hunk sessions exist, everything funnels through a session id. In practice you only care about this git checkout. Right now you either pick an id by hand, or pipe hunk session list --json through jq to match repoRoot. That’s awkward for humans and worse for automated agents.
Let hunk session subcommands (and ideally hunk session list) accept something like --repo path name and --worktree path name (defaulting to the current repo) so the CLI resolves the right session.
Same behavior people already get by filtering the session API; the CLI should make the common case one command.