Every command the replay CLI exposes. Run replay --help, or replay <command> --help, for the same information from your terminal.
Setup
replay init
First-run setup. Logs you in and optionally turns on background auto-upload. Idempotent — safe to re-run anytime to change your setup.
replay init
replay login
Authenticate with the server. Opens your browser, runs you through sign-in, then returns control to the CLI once the local callback completes.
replay login
replay logout
Remove the stored credentials on this machine.
replay logout
replay whoami
Show the currently authenticated account.
replay whoami
Browsing and searching
replay list
List the sessions discovered on this machine. Scoped to the current directory by default.
replay list # sessions under the current project
replay list --all # every session, across all projects
replay search
Search threads. Searches your uploaded threads by default; pass --local to search sessions on this machine instead. An optional query seeds the interactive picker, where you can refine live.
replay search "billing webhook" # uploaded threads
replay search "billing webhook" --local # local sessions, current project
replay search "billing webhook" --local --all # local, every project
--all only applies with --local — uploaded threads aren't project-scoped today.
Publishing
replay upload
Publish a local session. Pass a session id or exact title, or omit the argument to pick interactively. See /docs/upload-command for the full resolution and re-upload behavior.
replay upload # interactive picker
replay upload <session-id-or-title> # publish a specific session
| Flag | Effect |
|---|---|
-g, --global | Consider sessions across all projects, not just the current directory. |
--workspace <SLUG> | Publish into a workspace by slug, overriding your saved default. |
--personal | Publish to your personal account, ignoring any saved default workspace. |
--auto | Non-interactive sync used by replay auto. Uploads anything changed since its last push. Other flags are ignored when set. |
--workspace and --personal are mutually exclusive.
replay auto
Manage the background auto-upload schedule. When enabled, changed sessions sync on a timer (every 15 minutes) without you running replay upload. The schedule honors your ignore list and saved default workspace.
replay auto enable # install the schedule
replay auto disable # remove it; manual upload still works
replay auto status # whether it's scheduled, and when it last ran
replay workspace
List the workspaces you belong to and choose a default upload target.
replay workspace list # workspaces you can publish into
replay workspace use <slug> # set the default upload workspace
replay workspace use --personal # clear the default; uploads go to your personal account
Continuing a thread
replay continue
Resume a Replay thread on your local harness. Accepts a thread slug (hello-world-abc123) or a full thread URL. By default it resumes on the source's original agent; --with bounces it to a different one.
replay continue hello-world-abc123 # resume a thread by slug or URL
replay continue <session-id> --local # continue a local session by id
replay continue hello-world-abc123 --with codex # resume into a different agent
--with takes one of: claude, codex, opencode. --local treats the argument as a local session id (as printed by replay list) and skips the network round-trip.
Filtering
replay ignore
Manage the ignore list. Sessions whose project path sits under an ignored entry are filtered out of upload, list, and search.
replay ignore # print the current list
replay ignore <path> # add a path (shorthand for `add`)
replay ignore add <path> # add a path
replay ignore remove <path> # remove a path
replay ignore list # print the current list
Integrations
replay mcp
Run as an MCP server over stdio, so an AI assistant can search and read your sessions. See /docs/mcp-server for setup and the full tool list.
replay mcp
replay completions
Generate shell completions for your shell.
replay completions <shell> # e.g. bash, zsh, fish