Tableau Refresh Failure Agent with OpenCode
Two instructions carry this template and both are refusals: quote the error exactly, and never rerun anything. OpenCode treats the model as a setting rather than a given, so when a run starts paraphrasing Tableau's wording or offering to kick the schedule, the correction is a change of model rather than a rewrite of your escalation rules.
Watch it work before it's live
Run a staged conversation — no account needed. The agent handles it for real while a simulated world answers its tool calls; nothing touches real accounts, and nothing is actually sent.
Finance Daily Revenue
Extract refresh job 8f2c41 failed at 05:12 UTC after 41 seconds. Error: [Snowflake][ODBC] (39001) Incorrect username or password was specified. Unable to connect to the server. Scheduled daily at 05:00.
Set up in minutes
Using this template drops you into a guided setup. It asks exactly this, nothing else:
Connect Tableau
One sign-in. The agent acts through your account, scoped to what this template uses.
Connect Slack
One sign-in. The agent acts through your account, scoped to what this template uses.
Escalation rules
Which dashboards the business actually watches and who to tell when they go stale - your critical projects and the teams behind them, your quiet hours, and the first check you want suggested for credential failures and timeouts.
Runs on OpenCode
Preselected for this page — connect your OpenCode account during setup, or switch to NoClick's built-in models with one click.
Watch it handle a test run
A staged conversation against a simulated world — then it’s live.
Why OpenCode for this agent
Verbatim is a testable rule
You can compare the error string in Slack against the one in the job record, character for character. Very few instructions are that easy to audit after the fact.
Suggesting, not acting
The final section is a first step for a person, taken from your rules. A model that slides from suggesting into rerunning is the drift to test for, and it is a one-field change to correct.
Before you fork
Is a small model enough to write these notes?
For the quoting and the counting, usually, since neither asks much of the model. Where a smaller one slips is Error With No Detail, the staged failure whose message says nothing useful: the honest note admits the record is empty and falls back to the shared connection, while a weaker model reaches for a cause. Read that one before settling on a size.
Run it with a different agent
Put Tableau Refresh Failure Agent to work on OpenCode
Free to start. Guided setup, a test run against staged conversations, and it's live.