Agent Skills: Add Testing - Full TDD Pipeline

Add Testing - Full TDD Pipeline

Execute the complete TDD pipeline to add tests to an existing project or module.

Target path: $ARGUMENTS (if empty, use the current project directory)

Pipeline

Step 1: Testability Audit

Invoke the testability-auditor agent on the target path:

• Analyze coupling, global state, hidden dependencies
• Generate report with score 0-100
• Identify critical and moderate issues per file

Step 2: Conditional Code Adaptation

If the testability score is less than 80:

• Invoke the code-adapter agent with the report from Step 1
• Apply refactorings: interface extraction, dependency injection, I/O separation
• Verify compilation after each change
• Re-evaluate score if necessary

If the score is 80 or higher, skip to Step 3.

Step 3: Dependency Setup

Invoke the testing-deps-investigator agent:

• Detect project language and ecosystem
• Evaluate existing testing dependencies
• Recommend frameworks, assertion libraries, and coverage tools
• Research recommended testing workflow for the detected stack

Use AskUserQuestion to confirm which dependencies to install before proceeding.

Step 4: Test Implementation

Invoke the test-implementer agent module by module:

• Follow strict TDD: RED -> GREEN -> REFACTOR
• Run each test and confirm it fails before implementing
• Verify the entire suite passes after each implementation
• Cover: degenerate, simple, general, edge, and error cases

Consult ../../references/anti-patterns.md and ../../references/rg-refactor.md during implementation.

If the path contains multiple independent modules, process them in parallel when possible.

Step 4.5: Test Input Quality Audit

Invoke the test-input-auditor agent on all test files generated in Step 4:

• Score each file from 0 to 100
• Detect anti-patterns: The Liar, The Giant, tautological assertions, hardcoded mirroring, multiple behavioral assertions per test case
• Verify input category coverage per function: degenerate, simple, edge, and error cases

If any file scores below 80:

• Invoke test-implementer again targeting only those files
• Pass the test-input-auditor report as context so the implementer knows which input categories and anti-patterns to address
• After re-generation, invoke test-input-auditor again on the re-generated files only
• Repeat until all files score 80 or higher, or until two re-generation attempts have been made — whichever comes first

If all files score 80 or higher, proceed to Step 5.

Step 5: Coverage Report

Run the language's coverage tool and report:

• Total coverage for the project/module
• Coverage per file
• Files below 80%
• Number of tests created
• Summary of applied refactorings (if any)

Consult ../../references/coverage.md to interpret the report and identify files below the 80% threshold.

Step 6: Generate Project Testing Rules

Read ../../references/project-rules-template.md and generate .claude/rules/testing.md in the target project:

• Fill in the template with data discovered across Steps 1-5:
  • How to run tests and coverage (from Step 3 investigation)
  • Test organization and naming conventions (from Step 4 implementation)
  • Coverage target and exclusions (from Step 5 report)
  • Patterns established during the pipeline
• Create .claude/rules/ directory if it does not exist
• Report the file path created to the user

Reference Material

• Testing Anti-Patterns: 5 critical anti-patterns and the 3 iron laws
• Red-Green-Refactor Cycle: Detailed cycle with examples in Go, TypeScript, and C#
• Coverage Strategies: Tools per language, report interpretation, strategies for 80%+
• Project Rules Template: Template for generating .claude/rules/testing.md