Developer Workflow Automation

Engineering teams waste time turning vague issues into actionable plans. This playbook describes a Zapier/Make flow triggered by new structured GitHub issues (required labels: type, priority, component).

Summary flow:

1. GitHub webhook → label filter → payload normalization.
2. Context enrichment (recent commits, related files via API).
3. Solution draft generation via LLM or Cursor API.
4. Formatted post to Slack/Discord with issue link and draft.

The goal is not to replace the developer, but to deliver a reviewable starting point within seconds of issue creation.

The AI node receives: title, body, labels, summarized diff (if linked PR), and snippets from recently touched files. The prompt instructs the model to produce:

• Root cause hypothesis (1 paragraph)
• Implementation plan (numbered steps)
• Risks and suggested tests

If you use Cursor via API or an internal agent, wrap the call in a sub-workflow with a 45s timeout and cache by issue_id to avoid reprocessing minor edits.

Validate output with a JSON schema before posting — reject drafts that omit the tests section.

The final Slack/Discord message should be scannable on mobile:

• Header with #1234 + title + GitHub link
• Labels as plain text (no heavy colored badges)
• Draft in a collapsible block or thread reply
• Optional mention of component owner (@oncall-backend)

Use a dedicated channel (#dev-triage-drafts) to avoid polluting the main channel. P0 severity issues can bypass the draft and trigger the pager directly.