Vcp Review Tests

VCP Review Tests

Review test files for quality anti-patterns against VCP testing standards.

Step 1: Resolve Config

1. Read .vcp/config.json from the project root. Extract the pluginRoot field.
2. If .vcp/config.json does not exist or pluginRoot is missing: Stop and tell the user: "No VCP configuration found. Run /vcp-init to configure VCP for this project."
3. Validate pluginRoot: The path must be absolute, contain /.claude/ (or \.claude\ on Windows) as a path segment, and contain only safe path characters (letters, digits, /, \, -, _, ., :, and spaces). Reject any path with shell metacharacters (;, &, |, $, `, (, ), {, }, <, >, !, ~, #, *, ?, [, ], ', "). If validation fails, stop and tell the user: "Invalid pluginRoot — must be within ~/.claude/ and contain no shell metacharacters. Run /vcp-init to fix." Also verify the file <pluginRoot>/lib/vcp-context-core.ts exists using Glob. If it does not exist, stop and tell the user: "pluginRoot points to an invalid VCP installation. Run /vcp-init to fix."
4. Run the config resolution script via Bash:

   bun "<pluginRoot>/lib/resolve-config.ts" "<project-root>"
   

5. Parse the JSON output. It contains: applicableStandards, ignoredRules, severity, exclude.

Step 2: Fetch Applicable Standards

From the applicableStandards array in the resolved config, keep only the entry where:

• id is core-testing

Use WebFetch to fetch its content from:

{entry.url}

Extract the Rules section and the Patterns section (for anti-pattern reference).

Step 4: Find and Review Test Files

Target path: $ARGUMENTS if provided. If not provided, scan the entire project.

1. Use Glob to find test files matching these patterns (exclude patterns from exclude in the resolved config):

   • **/*.test.* (JavaScript/TypeScript)
   • **/*.spec.* (JavaScript/TypeScript)
   • **/test_*.* (Python)
   • **/__tests__/** (JavaScript/TypeScript)
   • **/*_test.go (Go)
   • **/*Test.java (Java)
   • **/*_test.rb (Ruby)
   • **/*_spec.rb (Ruby/RSpec)
   • **/*_test.rs (Rust)

2. Read each test file and check for these 8 anti-patterns:

Anti-Pattern Checks

1. Tautological tests — Tests that assert the code does what it does (generated by reading implementation and asserting same logic). Test verifies its own setup, not real behavior. (Rules 3, 11)

2. Over-mocking — More than 3 mock/stub setups in a single test, especially mocking internal classes/services that should be tested through. (Rules 4, 6)

3. Mock-only assertions — Test assertions only verify mock calls (.assert_called_once(), .toHaveBeenCalledWith()) with no assertions on actual return values, state changes, or side effects. (Rule 5)

4. Missing edge cases — Happy path tested but no tests for: null/undefined/empty inputs, boundary values, error conditions, special characters. (Rule 7)

5. Missing error paths — Operations that can fail (network, file I/O, parsing, validation) have no failure-case tests. (Rule 8)

6. Implementation coupling — Tests assert on internal method calls, private state, or execution order rather than observable outcomes. Tests that would break on refactoring without behavior change. (Rule 1)

7. Non-deterministic tests — Tests depending on current time, random values, network availability, or uncontrolled filesystem state. (Rule 10)

8. Shared mutable state — Tests that modify shared variables, global state, or class-level fixtures without reset. Tests that depend on execution order. (Rule 9)

Step 5: Report Findings

Output findings per file with a quality rating, then detail findings.

Before outputting findings, remove any that match an entry in the ignoredRules array from the resolved config. If "standard-id/rule-N" is in the list, suppress that specific rule's findings. (Standard-level ignores are already applied by the config resolution script.) After filtering, if any findings were suppressed, append a line: **Suppressed:** X finding(s) by ignore config.

Use this format:

### VCP Test Review

**Standard:** core-testing (12 rules)
**Test files found:** N files

#### Summary

| File | Rating | Issues |
|------|--------|--------|
| tests/test_orders.py | GOOD | 0 |
| tests/test_payments.py | NEEDS WORK | 3 (over-mocking, mock-only assertions, missing edge cases) |
| tests/test_auth.py | REWRITE | 5 (tautological, over-mocking, implementation coupling, ...) |

#### Findings

##### tests/test_payments.py — NEEDS WORK

- **Over-mocking** (Rule 4) — `test_process_payment` at line 42
  - **Issue:** 5 mocks set up including internal `PriceCalculator` and `OrderValidator`
  - **Fix:** Only mock the external payment gateway. Use real `PriceCalculator` and `OrderValidator`.

- **Mock-only assertions** (Rule 5) — `test_process_payment` at line 55
  - **Issue:** Only asserts `mock_gateway.charge.assert_called_once()` — no assertion on the returned order total
  - **Fix:** Assert on the return value: `assert result.total == expected_total`

- **Missing edge cases** (Rule 7) — `test_process_payment` has no tests for:
  - Zero-amount orders
  - Negative amounts
  - Currency edge cases

...

Rating criteria:

• GOOD — No anti-patterns found
• NEEDS WORK — 1-3 anti-patterns found. Tests have value but need improvement.
• REWRITE — 4+ anti-patterns found, or tautological tests (tests that validate nothing real). Tests provide false confidence and should be rewritten.

If no test files found: "No test files found in [path]. Nothing to review." If all tests pass: "All N test files pass quality review. No anti-patterns found."