Skip to content
CosmosCLIChangelog
Context EnginePricingDocsBlog
CustomersCareersStatus PageTrust CenterSecurity
Talk to SalesSign in
Book demo

Coding

Slack feedback to merged PR

Turn Slack feedback into a tracked fix: triage the message, open a ticket, implement the change, and reply when the PR merges.

slackfeedbacklinearjirapull requestautomationtriagecodingfull loop

[ workflow / coding ]

Slack feedback to merged PR

Feedback lands in a designated Slack channel. Cosmos classifies it as bug, enhancement, docs gap, or noise, enriches it with codebase context, and opens a Linear or Jira ticket when action is needed. An implementation agent writes the fix, opens the PR, and replies in the original thread after merge.

14 nodes

10 edges

Trigger[trigger]
Feedback posted in Slack

Designated feedback channel

AI Agent step[classify]
Classify message

Bug / enhancement / docs / noise

Decision

Actionable feedback?

Not noise or off-topic

No
Bypass (already solved)[react-skip]
React + skip

Emoji ack, no ticket

YES
AI Agent step[enrich]
Enrich with codebase context

Affected area + related tickets

Decision

Duplicate ticket?

Open ticket for same area

Yes
Bypass (already solved)[link-existing]
Link to existing ticket

Notify reporter + end

YES
Output / Result[open-ticket]
Open Linear / Jira ticket

Title, context, severity

AI Agent step[implement]
Implement fix on branch

Code + tests + self-review

Output / Result[open-pr]
Open pull request

Ticket ref + checklist

Human-in-the-loop[human-review]
Maintainer reviews PR

Approve or request changes

Decision

PR merged?

Approved and shipped

No
AI Agent step[revise]
Iterate on feedback

Loop back to review

YES
Output / Result[notify]
Reply in Slack thread

Shipped confirmation + link

Workflow prompt

Paste this into Augment to reproduce the workflow end-to-end.

Build a Cosmos workflow that turns a Slack feedback message into a merged pull request with no human filing anything.

Trigger: a new message is posted in a designated feedback Slack channel (e.g. #feedback-billing, #feedback-search).

Steps:
1. Classify the message. Determine whether it is a genuine bug, a feature/enhancement request, a documentation gap, or noise (thank-you, off-topic, duplicate intent). If it is noise, react with an emoji and end.
2. Enrich the feedback. Search the codebase and recent commit history for the area of code most likely responsible. Pull any related existing tickets that are already open or recently closed for the same area.
3. Decision: "Duplicate of an existing open ticket?".
   - If yes, link the Slack message to the existing ticket, notify the reporter, and end.
   - If no, continue.
4. Open a ticket in Linear or Jira with a clear title, a description derived from the feedback message, the affected area, the enriched codebase context, and a link back to the Slack message. Assign a severity and initial priority.
5. Assign the ticket to an implementation agent (or the relevant team queue). The agent creates a new branch, implements the fix or enhancement, writes tests, and self-reviews the diff.
6. The agent opens a pull request against the main branch, with the ticket reference in the title, a summary of what changed, and a checklist of the acceptance criteria from the ticket.
7. A human maintainer reviews the PR. They can request changes; the agent iterates.
8. Decision: "PR approved and merged?".
   - If no, loop back to the revision step.
   - If yes, continue.
9. Post a reply in the original Slack thread: "Shipped in [PR link]: [one-sentence summary of the change]." React with ✅.

Constraints:
- Never implement a change that touches auth, billing, or data-deletion paths without a human review step flagged explicitly in the PR.
- Always link ticket ↔ PR ↔ Slack message so the chain of evidence is traceable.
- Keep the feedback triage log append-only for trend analysis (volume, category, area, cycle time).
← All Workflows