---
name: "test"
description: "Enter the Test phase of CocoBrew. Reads spec.md test requirements, generates test cases, executes SQL validation and quality checks, records results in test.md. Can be re-run without full rebuild. Requires Build phase completion."
version: "1.0.0"
author: "CocoPlus"
tags:
  - cocoplus
  - lifecycle-engine
---

You are executing the Test phase (4/6) of CocoBrew.

Before proceeding, verify that `.cocoplus/` exists.
If not: output "CocoPlus not initialized in this directory. Run `$pod init` to begin." Then stop.

Read `.cocoplus/lifecycle/meta.json`. Verify `phases_completed` contains `"build"`.
If not: output "The Build phase must be completed before testing. Run `$build` first." Then stop.

## Generate Test Cases

Read `.cocoplus/lifecycle/spec.md`. For each Success Criterion and Deliverable:
1. Generate a test case that validates that criterion or deliverable exists
2. Each test case has: ID, description, test approach (SQL query / file existence / quality check), expected result

## Execute Tests

For each test case:
1. Run the test (SQL validation via SnowflakeSqlExecute if applicable, file existence check via Bash, quality check via quality-advisor)
2. Record: PASS or FAIL with actual result
3. If a test initially fails because of environment setup, permissions, missing metadata, or transient service behavior, document the recovery step explicitly before re-running. Do not hide the failed attempt.

Recovered tests count as PASS only when the final assertion actually passes. The original failure must still appear under `## Recoveries` with the failed command/query, error, recovery action, and final outcome.

## Write Test Results

Write `.cocoplus/lifecycle/test.md`:

```markdown
# Test Results

**Date:** [ISO 8601 timestamp]
**Phase:** Test (4/6)
**Phase ID:** test-YYYYMMDD-001

## Summary
- Tests Run: [count]
- Passed: [count]
- Failed: [count]

## Test Cases

### [TEST-001]: [description]
**Approach:** [SQL / file check / quality]
**Expected:** [expected result]
**Actual:** [actual result]
**Status:** PASS / FAIL

[repeat for each test case]

## Failures
[Detail any failures with remediation suggestions]

## Recoveries
[For each recovered test: original command/query, error observed, recovery action taken, final PASS/FAIL]
```

## Update State

Update `lifecycle/meta.json` — add test to phases_completed.
Append to AGENTS.md (≤200 lines): `Phase: Test (4/6) — [PASS/FAIL summary]`

## Git Commit

```
git add .cocoplus/lifecycle/test.md .cocoplus/lifecycle/meta.json .cocoplus/AGENTS.md
git commit -m "test: test execution and validation"
```

Output: "Test phase complete. [N] tests passed, [M] failed. Check `.cocoplus/lifecycle/test.md` for details. Proceed to `$review`."

## Anti-Rationalization

| Shortcut / Temptation | Why It Fails |
|-----------------------|--------------|
| Skip tests for deliverables that "obviously exist" | The spec mandates verifiable outcomes — assumption-based passing is not evidence |
| Mark phase complete even if some tests failed | Failing tests are open issues; committing and moving on hides defects from the Review phase |
| Generate test cases without reading spec.md success criteria | Test cases not anchored to success criteria will miss spec-mandated outcomes entirely |
| Collapse recovered failures into clean passes | Recovery behavior is release evidence; hiding it removes the signal needed to fix flaky or environment-sensitive tests |

## Exit Criteria

- [ ] `.cocoplus/lifecycle/test.md` exists with a `## Summary` section showing Tests Run, Passed, and Failed counts
- [ ] `.cocoplus/lifecycle/test.md` contains `## Recoveries`, even if it says "No recovered failures"
- [ ] Every Success Criterion from `spec.md` has at least one corresponding test case in `test.md`
- [ ] `.cocoplus/lifecycle/meta.json` `phases_completed` array contains `"test"`
- [ ] Git commit with message `test: test execution and validation` exists in log