Skip to main content

How it works

The ready-made Feedback Triager uses Slack: it responds to new top-level messages in a configured channel, investigates reports, checks for duplicates, and follows the thread. It creates tickets automatically when quality checks pass, or only on request, according to your settings. Otherwise, it answers or asks for missing evidence. This diagram shows the Slack setup: Incident Investigator supports two Slack-based modes: alert-thread mode investigates new reports in configured alert channels, and stationed-channel mode follows an incident channel. In both modes, code fixes require explicit human approval before it launches PR Author; ticket creation is not automatic.

Where you participate

  • Choose the filing policy. Select automatic or on-request tickets, the tracker destination, and routing rules.
  • Provide context. Supply expected behavior, a reproduction, and missing evidence. Review findings separately from hypotheses.
  • Authorize implementation. Explicitly request a specific code fix or delegate a ticket. Ticket creation alone is not approval to implement. Investigation does not authorize production changes.

Template Experts

  • Feedback Triager: investigates Slack feedback and files in Linear, Jira, or GitHub Issues under your policy; it does not edit code.
  • Incident Investigator: provides incident findings, recommended actions, and follow-up.
  • PR Author: implements a human-requested fix and carries it into review.
  • Ticket Manager: coordinates work explicitly delegated from the tracker.

Set it up

For the ready-made Slack setup:
  1. Connect Slack, your tracker, and code host. Make sure the Slack app can access the channel and the Experts can access your repository and evidence sources.
  2. Configure Feedback Triager with Cosmos Advisor. Advisor configures the channel, ticket destination, filing policy, repository, and required Author worker. That worker still needs your request before starting a fix.
  3. Complete setup and approve activation. Post a top-level report in the configured channel. Inspect the thread and automation run history for evidence, duplicate handling, and the next step.
Example Advisor prompt:
Set up Feedback Triager for [Slack channel] and [ticket tracker]. Create tickets [automatically / only when requested].

Next step

Delegate ready work to Ticket → PR, or take broader needs to Requirements → Ticket. Code fixes continue through PR → Merge.

Further reading