From creating a ticket to production deploy — what happens automatically, what you do, and where to look.
Create issues, trigger builds, review QA artifacts, iterate until ready for stakeholders.
Where you work: Claude Code CLI (a terminal, inside a cloned repo). Triggering Sling is a /sling-prep command. Producer workflow →
Review staging, provide feedback in Slack, approve or request changes. Push to production when ready.
Where you work: Slack — for reviewing finished work. You don’t start work from Slack. Owner workflow →
Set up repos, manage GitHub Projects, monitor deploys, handle rollbacks if needed.
Where you work: config/repositories.json, server SSH, GitHub branch protection.
If you only remember three commands:
gh repo clone wbtw-repositories/<repo> ~/Code/<repo> # one-time per repo
cd ~/Code/<repo> && claude # open Claude Code there
/sling-prep https://github.com/wbtw-repositories/<repo>/issues/<n>Green = Sling does it automatically. Blue = you do it. Orange = Slack interaction.
/sling-prep <issue-url>. Fallback: add the auto-implement + claude-code labels directly on GitHub.The whole trigger sequence in 3 commands:
gh repo clone wbtw-repositories/<repo> ~/Code/<repo> # one-time per repo
cd ~/Code/<repo> && claude # open Claude Code in that folder
/sling-prep https://github.com/wbtw-repositories/<repo>/issues/<n>That’s it. Claude walks you through prep, applies the labels, Sling picks up within ~2 minutes, and you’ll get a Slack ping when staging is ready.
Write clear requirements as a GitHub issue in the target repo. Include:
Recommended — in Claude Code:
/sling-prep https://github.com/owner/repo/issues/123Claude scores the issue, suggests improvements, and applies the right labels for you. This is the fastest path and catches scope problems before Sling runs.
Manual fallback — on GitHub:
Add the auto-implement + claude-code labels directly. Sling picks it up via webhook within ~1–2 minutes.
Coming soon: Slack-based trigger via the AIA bot. Not live yet — use Claude Code or GitHub for now.
Sling will:
Average time: 15–45 minutes per issue.
The Slack notification includes:
If changes are needed before showing to the business owner, create a new issue or update the existing one with additional requirements. Re-run /sling-prep or add auto-implement again.
Tip: Be specific in change requests. “Make the button blue” works better than “improve the UI.”
This page is only for reviewing finished work. If you’re trying to start a new build, you’re on the wrong page — go to For Product Producers instead. Producers trigger Sling from Claude Code (the CLI). Slack only delivers the demo-ready ping after Sling finishes.
Once a producer has triggered Sling and it has built + staged the work, as a business owner your review happens in Slack — approve, request changes, or trigger a rollback. As a reviewer you never need to touch GitHub or the codebase.
You’ve tested on staging and it’s good. Click Approve. Your approval is recorded. Once the threshold is met (e.g., 1 of 1, or 2 of 3), production push is unlocked.
Something’s wrong. Reply in the Slack thread with your feedback, then click Request Changes. Your feedback is posted to GitHub and the team is notified. All approvals are reset.
Only active after approval threshold is met. Click it to deploy. Sling merges the PR and monitors the deploy. You’ll see a success message with a Rollback button.
Multiple business owners can review the same feature. Each person’s approval counts toward the threshold. Any single “Request Changes” resets everyone’s approvals.
Every task moves through these columns automatically. No manual updates needed.
The GitHub Project board can be made public so stakeholders see task progress without accessing the repository code. Ask an engineer to set this up.
If something goes wrong after deploying to production, you can revert in one click.
git revert commitRollback reverts code only. Database migrations (e.g., php artisan migrate) are NOT automatically reversed.
If the original PR included migration files, Sling posts a warning before proceeding.
If you need to reverse a migration, contact an engineer. Do not attempt manual database changes.
Sling has two execution modes. The right one is chosen automatically based on repo config or issue labels.
| Mode | How It Works | Best For | How to Trigger |
|---|---|---|---|
| STIV (default) | Spec → Test → Implement → Validate. Uses OpenRouter API. Each phase is tracked separately. | Well-scoped issues, single-file changes, TDD-friendly tasks | Add auto-implement label (no extra label needed) |
| Claude Code | Headless Claude Code CLI execution. Full codebase context. Uses Max subscription (flat-rate). | Complex multi-file changes, refactors, tasks needing broad context | Add claude-code label to the issue |
Start with the default (STIV). If it fails or produces incomplete results, re-create the issue with the claude-code label for the more powerful mode. Most Diagnostic.ly repos default to claude-code already.
Three things happen automatically after every successful production deploy.
Claude generates a “reproduction prompt” from the task + PR diff. It’s committed to a knowledge repo on GitHub so the team can learn from and recreate past work.
The PR diff is submitted to sctxt.ai for compliance scoring against Diagnostic.ly governance standards. Results posted to Slack + GitHub PR.
Claude extracts reusable engineering guidelines from the deploy and submits them as proposals to sctxt.ai. An admin reviews and approves them as team standards.
| I Want To… | Do This |
|---|---|
| Start a new task | Create GitHub issue → add auto-implement label |
| Use Claude Code mode | Also add the claude-code label to the issue |
| Check task status | Look at the GitHub Projects board, or the Sling dashboard at sling.svenlabs.ai/dashboard |
| Review a build | Click the staging URL in the Slack notification |
| Give feedback | Reply in the Slack thread, then click Request Changes |
| Approve for production | Click Approve in Slack |
| Deploy to production | Click Push to Production (after approval threshold met) |
| Roll back a deploy | Click Rollback on the deploy success message in Slack |
| See what’s in production | “In Production” column on the GitHub Projects board |
| Talk to Sling via Slack | Message AIA in Slack: status, approve, pending |
| Dashboard | sling.svenlabs.ai/dashboard |
| Documentation | sling.svenlabs.ai/docs |
| Implementation Report | sling.svenlabs.ai/workflow |
| Status Page | sling.svenlabs.ai/status-page |
| app.diagnostic.ly | UAT enabled |
| diagnostic.ly-nodejs | UAT enabled |
| laravel.diagnostic.ly | UAT enabled |
| whmgr-node | STIV only |
| whmgr-laravel | STIV only |
| workos | STIV only |