Skip to content

bootstrap-test.md

Latest commit

 

History

History
112 lines (79 loc) · 4.41 KB

bootstrap-test.md

File metadata and controls

112 lines (79 loc) · 4.41 KB
Raw

Bootstrap Test — Agent Swarm Validation

Run this sequence after Slack, Cursor integrations, and all automations are enabled.

Prerequisites

Test sequence

Post each message as a new top-level message in the listed channel. Expect a threaded reply and ✅ on your message when the run completes.

Step 1 — Chief of Staff (#vacation-hq)

We want to improve phone-number handoff reliability.

Expected: Agent breaks down the work and tells you to post follow-ups in #vacation-product and/or #vacation-engineering. It should not write code.

Step 2 — Product (#vacation-product)

Review lead_extraction_heuristic.dart and propose one change to reduce premature handoffs before we have destination and budget.

Expected: Specific proposal referencing flutter_app/lib/chat/lead_extraction_heuristic.dart and tradeoffs. If files change: automatic commit, push, Firebase deploy, production URL in thread.

Step 3 — Engineering (#vacation-engineering)

Open a draft PR that adds a one-line comment above shouldHandOff in lead_extraction_heuristic.dart explaining the handoff rules. Do not change behavior.

Expected: Change committed to main, pushed, Firebase hosting deployed automatically; thread summary with file paths, commit hash, deploy result, and production URL.

Step 4 — QA (#vacation-qa)

Run flutter analyze and the test scripts from AGENTS.md. Report results.

Expected: Pass/fail report with commands run. When tests pass: automatic commit, push, build-web.ps1, Firebase deploy — no permission prompt. Failures should include repro steps.

Step 5 — Deploy (#vacation-deploy)

Run the AGENTS.md release pipeline now — build and Firebase deploy hosting.

Expected: Agent automatically commits (if needed), pushes main, runs build-web.ps1, deploys Firebase hosting (and functions if changed), posts deploy result and production URL — never asks for permission, never skips deploy.

Step 6 — Ops (#vacation-ops)

What happens when submitLead succeeds but Twilio SMS fails? Point me to the relevant code paths.

Expected: References functions/src/index.ts and Firestore/SMS flow without exposing secrets.

Step 7 — PR Review Bot (GitHub)

Open any test PR on ether314/Vacation (can be draft-free trivial change).

Expected: PR comment from bot + summary post in #vacation-engineering.

Pass criteria

Step Pass
1 Delegation to correct channels
2 Product-specific file references; auto Firebase deploy if code changed
3 Commit, push, automatic Firebase deploy, thread summary
4 Test commands executed; auto deploy when pass; clear env blocker if credentials missing
5 Automatic Firebase deploy without prompting — never skip
6 Accurate ops guidance, no secrets
7 Automated PR review on new PRs

If something fails

Symptom Fix
No agent response Automation disabled, wrong channel selected, or Slack not connected
"Cloud compute required" Enable Cloud Agents in Cursor dashboard
Channel not in picker Re-authorize Slack integration; confirm channel is public
Tests fail on credentials Expected locally — QA agent should report blocker, not fake pass
PR bot silent Confirm GitHub integration and ether314/Vacation repo scope on trigger

Repo validation script

From repo root (checks swarm artifacts exist):

$files = @(
  "AGENTS.md",
  "docs/slack-workspace-setup.md",
  "docs/cursor-integrations-setup.md",
  "docs/bootstrap-test.md",
  "docs/automations/chief-of-staff.json",
  "docs/automations/product-agent.json",
  "docs/automations/engineering-agent.json",
  "docs/automations/qa-agent.json",
  "docs/automations/deploy-agent.json",
  "docs/automations/ops-agent.json",
  "docs/automations/pr-review-bot.json"
)
$missing = $files | Where-Object { -not (Test-Path $_) }
if ($missing) { Write-Error "Missing: $($missing -join ', ')" } else { Write-Host "All swarm artifacts present." }