What does tares status check, and what does it mean when it exits with code 1?
The tares status command in Tares asks the running daemon and prints one line per readiness check, ending with the one thing to do next. Exit code 1 means nothing answers, so no Tares daemon replied at the address the command tried. Add --json to get the same result as a stable object.
How it works
tares status asks the daemon, so it reports what the running Tares sees, not what a config file says. It finds the daemon in this order:
- the
--taresdflag - the
TARESD_URLenvironment variable - the workspace that
tares loginsaved - the local Tares at
http://127.0.0.1:8787
It prints one line per check:
- daemon: whether the daemon answers.
- auth: whether a login is required.
- sources: how many are configured and whether they are receiving.
- projects: the projects in this Tares.
- triggers: the triggers in this Tares.
- Tares agents: the agents, and whether an Anthropic key is set.
- MCP endpoint: the endpoint agents connect to.
- agent clients: which clients on this machine have Tares registered. It checks Claude Code, Cursor and Codex, and only reads, so it never changes their config.
- Slack: the Slack check.
The last line is always the one thing to do next, so you can read the output from the bottom up.
When you are signed in with tares login, the first line says which workspace you are talking to.
Exit code 1 means nothing answers. The daemon is not running, or the address you gave, or the one found by the lookup order above, points at nothing that replies.
When auth is on, the command needs the root token to ask the daemon. It reads the token from --data-dir (the directory tares up used) or from the TARES_AUTH_TOKEN environment variable.
Set it up
- Run the check against the default local Tares:
tares status-
Read the last line. It names the one thing to do next, so do that first and run the command again.
-
If the exit code is 1, start the daemon with
tares up, which serves on127.0.0.1port 8787 by default, then runtares statusagain.
tares up-
If you started Tares somewhere else, point the command at it with
--taresdfollowed by the daemon URL, or setTARESD_URL. A machine signed in withtares loginchecks its saved workspace first. To check a localtares upfrom such a machine, setTARESD_URLtohttp://127.0.0.1:8787or runtares logout. -
If your Tares runs with auth on, give the command the root token. Pass
--data-dirwith the directory you started Tares with, or exportTARES_AUTH_TOKEN. Bare--authontares upgenerates and persists a root token, and--auth=TOKENuses your own. -
For scripts and dashboards, use
--json. It prints the same checks as a stable object, so you can parse it instead of reading lines.
The full flag list is on the CLI reference.
What you get
- One readiness checklist for the whole setup: daemon, auth, sources, projects, triggers, agents, MCP endpoint, agent clients and Slack.
- A single next step on the last line, so you always know what to do before anything else.
- A quick answer to whether your agents have what they need, because the Tares agents line shows whether an Anthropic key is set.
- A view of which of Claude Code, Cursor and Codex on this machine already have Tares registered, without touching their config.
- A stable
--jsonobject you can use in scripts. - A clear failure signal: exit code 1 tells your script or CI step that nothing answered.
If you run Tares for monitoring and incident response, as described in AI SRE, run tares status after a restart or a config change. It shows whether sources are receiving and whether triggers and agents are in place.
When something else fits
tares statusis a readiness check, not a log viewer. To see what a source receives or what a trigger did, use the console thattares upprints.- To check what the software factory for Claude Code needs, run
tares factory doctor, which checks everything the crew needs. - To change how a command finds Tares, such as signing in to a workspace or removing the saved login, use
tares loginandtares logout.
Related
Still stuck? Get help