Skip to content

Latest commit

 

History

History

e2e

README.md

E2E tests for Moltworker

End-to-end tests that deploy real Moltworker instances to Cloudflare infrastructure.

Why cloud-based e2e tests?

These tests run against actual Cloudflare infrastructure—the same environment users get when they deploy Moltworker themselves. This catches issues that local testing can't:

  • R2 bucket mounting only works in production (not with wrangler dev)
  • Container cold starts and sandbox behavior
  • Cloudflare Access authentication flows
  • Real network latency and timeout handling

Architecture

┌─────────────────────────────────────────────────────────────────────────┐
│                              Test runner                                │
│                                                                         │
│   cctr test/e2e/                                                        │
│     ├── _setup.txt      (start server, browser, video)                  │
│     ├── pairing_and_conversation.txt                                    │
│     └── _teardown.txt   (stop everything, clean up cloud resources)     │
└─────────────────────────────────────────────────────────────────────────┘
                                    │
                                    ▼
┌─────────────────────────────────────────────────────────────────────────┐
│                        Cloud infrastructure                             │
│                                                                         │
│   Terraform (main.tf)           Wrangler deploy           Access API    │
│   ├── Service token      →      ├── Worker           →    ├── App       │
│   └── R2 bucket                 ├── Container             └── Policies  │
│                                 └── Secrets                             │
└─────────────────────────────────────────────────────────────────────────┘
                                    │
                                    ▼
┌─────────────────────────────────────────────────────────────────────────┐
│                         Deployed worker                                 │
│                                                                         │
│   https://moltbot-sandbox-e2e-{id}.{subdomain}.workers.dev              │
│                                                                         │
│   Protected by Cloudflare Access:                                       │
│   - Service token (for automated tests)                                 │
│   - @cloudflare.com emails (for manual debugging)                       │
└─────────────────────────────────────────────────────────────────────────┘

Test flow

  1. Terraform creates isolated resources: service token + R2 bucket
  2. Wrangler deploys worker with unique name (timestamp + random suffix)
  3. Access API creates Access application (must be after worker exists—workers.dev domains require the worker to exist first)
  4. plwr opens browser with Access headers, navigates to worker
  5. Tests run with video recording capturing the full UI flow
  6. Teardown deletes everything: Access app → worker → R2 bucket → service token

Key design decisions

  • Unique IDs per test run: $(date +%s)-$(openssl rand -hex 4) ensures parallel test runs don't conflict
  • Access created post-deploy: Terraform can't create Access apps for non-existent domains
  • Container names: Derived from worker name as {worker-name}-sandbox

Test framework: cctr + plwr

Tests use two complementary tools:

cctr - CLI Corpus Test Runner

cctr runs tests where each test case is a command line script, e.g.

===
navigate to admin page to approve device
%require
===
TOKEN=$(cat "$CCTR_FIXTURE_DIR/gateway-token.txt")
WORKER_URL=$(cat "$CCTR_FIXTURE_DIR/worker-url.txt")
plwr -S moltworker-e2e open "$WORKER_URL/_admin/?token=$TOKEN"
---

Key features:

  • Plain text format: Easy to read and write
  • %require directive: If this test fails, skip all subsequent tests
  • Variables: Capture dynamic output with {{ name }}
  • Fixtures: fixture/ directory copied to temp dir for each suite
  • Setup/teardown: _setup.txt and _teardown.txt run before/after tests

plwr - Browser automation CLI

plwr provides shell-friendly browser automation with CSS selectors:

plwr -S test start
plwr -S test open "https://example.com"
plwr -S test wait 'text=Hello'
plwr -S test click 'button:has-text("Submit")'
plwr -S test fill textarea 'Hello world'
plwr -S test video-start
plwr -S test screenshot
plwr -S test stop

Example test

Here's a complete test that approves a device and sends a chat message:

===
wait for Approve All button and click it
%require
===
plwr -S moltworker-e2e click 'button:has-text("Approve All")' -T 120000
---

===
wait for approval to complete
%require
===
plwr -S moltworker-e2e wait 'text=No pending pairing requests' -T 120000
---

===
type math question into chat
%require
===
plwr -S moltworker-e2e fill textarea 'What is 847293 + 651824? Reply with just the number.'
---

===
wait for response containing the correct answer
===
plwr -S moltworker-e2e wait 'text=1499117' -T 120000
---

Running the e2e test suite locally

Prerequisites

  1. Copy .dev.vars.example to .dev.vars and fill in credentials (see file for detailed instructions)
  2. Install dependencies: npm install
  3. Install cctr: brew install andreasjansson/tap/cctr or cargo install cctr
  4. Install plwr: see plwr install instructions
  5. Install Playwright browsers: npx playwright install chromium

Run tests

# Run all e2e tests
cctr test/e2e/

# Run with verbose output
cctr test/e2e/ -v

# Run specific test file
cctr test/e2e/ -p pairing

# Watch test output in real-time (for debugging)
cctr test/e2e/ -vv

Run headed (see the browser)

PLAYWRIGHT_HEADED=1 cctr test/e2e/

View test videos

Videos are saved to /tmp/moltworker-e2e-videos/ after each run.