Example:

/test-design

I implemented a new slash command `/export` that exports chat history to Markdown.

Changed files:
- lua/vibing/application/chat/slash_commands.lua (added export_command)
- lua/vibing/utils/markdown_exporter.lua (new file)

Existing tests:
- tests/e2e/chat_basic_flow_spec.lua (basic chat operations)

Output Format

1. Test Scenario Analysis

Categorized scenarios with checkboxes:

Happy Path ✅ - Most common use cases
Error Cases ❌ - Error handling (validation, network, permissions)
Edge Cases 🔸 - Boundary conditions, special characters, timeouts
Integration Points 🔗 - Interactions with other features

2. Priority Ranking

Critical: Core functionality - must work for release
High: Error handling - security, data loss prevention
Medium: Usability - edge cases, user experience
Low: Performance - optimization, rare scenarios

3. Test Code Templates

Ready-to-implement code using e2e_helper.lua:

describe("E2E: [Feature Name]", function()
  local nvim_instance

  before_each(function()
    nvim_instance = helper.spawn_nvim_instance({
      headless = true,
      init_script = "tests/minimal_init.lua",
    })
  end)

  after_each(function()
    helper.cleanup_instance(nvim_instance)
  end)

  -- Test cases with appropriate TIMEOUTS constants
end)

Workflow

Gather context - Read changed files, analyze scope, review existing tests
Design scenarios - Generate Happy/Error/Edge/Integration cases
Prioritize - Rank by Critical/High/Medium/Low
Generate code - Create implementation-ready test templates

Example Output

For a /export command that exports chat to Markdown:

Test Scenarios

Happy Path ✅

Export chat with single message
Export chat with multiple messages
Export with frontmatter preserved

Error Cases ❌

Permission denied (read-only directory)
Invalid file path (path traversal attack)
Empty chat (no messages)

Edge Cases 🔸

Very long chat (1000+ messages)
Special characters in messages

Integration Points 🔗

Export from worktree chat
Export during active Assistant response

Priority Ranking

Critical

Export basic chat successfully
Handle permission errors gracefully

High

Validate file path (security)
Handle empty chat

Medium

Export with special characters

Test Code

-- tests/e2e/export_command_spec.lua
local helper = require("vibing.testing.e2e_helper")

local TIMEOUTS = {
  CHAT_CREATION = 2000,
  COMMAND_EXECUTION = 3000,
}

describe("E2E: /export command", function()
  local nvim_instance

  before_each(function()
    nvim_instance = helper.spawn_nvim_instance({
      headless = true,
      init_script = "tests/minimal_init.lua",
    })
  end)

  after_each(function()
    helper.cleanup_instance(nvim_instance)
  end)

  it("should export chat to markdown file", function()
    helper.send_keys(nvim_instance, ":VibingChat<CR>")
    vim.wait(TIMEOUTS.CHAT_CREATION)

    helper.send_keys(nvim_instance, "GiTest message<Esc><CR>")
    local ok = helper.wait_for_buffer_content(
      nvim_instance,
      "## .* Assistant",
      30000
    )
    assert.is_true(ok, "Assistant should respond")

    -- Execute /export and verify...
  end)
end)

Best Practices

Do:

Provide detailed feature description with expected behavior
List all changed files (git diff --name-only)
Review generated scenarios - don't blindly accept
Implement Critical/High tests first

Don't:

Use vague descriptions ("Added some features")
Skip review - templates need customization
Implement all tests at once - prioritize

Next Steps

After running /test-design:

Review generated test scenarios
Approve Critical and High priority tests
Implement approved tests in tests/e2e/
Run npm run test:e2e
Apply 3-try auto-fix rule if failures occur (see .claude/rules/self-testing.md)

References

.claude/rules/self-testing.md - Complete self-testing procedures
lua/vibing/testing/e2e_helper.lua - E2E helper API reference
tests/e2e/chat_basic_flow_spec.lua - Example test suite

test-design

Test Design for vibing.nvim

Usage