memoriant-test-coverage-skill

A Claude Code plugin that finds every untested function in a Python project using AST inspection and generates ready-to-fill pytest skeleton files for each gap.

Built on the methodology of NathanMaine/semantic-test-coverage-agent and packaged as a portable Claude Code skill + agent.

---

What It Does

1. Scans src/ recursively using Python's ast module — no mutation testing, no heuristic scoring.
2. Maps every public function and method to existing test_* functions in tests/ using naming-convention matching.
3. Produces a structured JSON gap report listing every uncovered function.
4. Generates a tests/test_<module>.py skeleton file for each source module with gaps — complete with imports, one class Test<Name> per function, and a _happy_path / _edge_case stub pair.
5. Never overwrites existing tests. Never touches source code.

---

Installation

Claude Code

claude mcp add memoriant-test-coverage-skill

Or clone and reference locally:

git clone https://github.com/NathanMaine/memoriant-test-coverage-skill ~/.claude/plugins/memoriant-test-coverage-skill

Codex CLI

codex install NathanMaine/memoriant-test-coverage-skill

Gemini CLI

gemini extension install NathanMaine/memoriant-test-coverage-skill

---

Usage

In Claude Code (natural language)

Find what I'm missing in my test suite

Generate pytest stubs for all untested functions in src/

Run semantic test coverage on ./backend and write skeletons to ./backend/tests

Via skill invocation

/semantic-test-coverage --project-path ./my_project --output-dir ./my_project/tests

Via Codex CLI

codex run test-coverage-agent --var project_path=./my_project

---

Plugin Structure

memoriant-test-coverage-skill/
├── .claude-plugin/
│   └── plugin.json              # Plugin manifest
├── skills/
│   └── semantic-test-coverage/
│       └── SKILL.md             # Full methodology for Claude Code
├── agents/
│   └── test-coverage-agent.md  # Autonomous agent definition
├── AGENTS.md                    # Codex CLI agent definitions
├── gemini-extension.json        # Gemini CLI extension manifest
├── SECURITY.md                  # Security policy
├── README.md                    # This file
└── LICENSE                      # MIT

---

Configuration

Parameter	Default	Description
project_path	.	Root of the Python project to scan
output_dir	tests/	Where to write generated skeleton files
include_private	false	Also generate stubs for _private functions
format	json	Gap report format: json or markdown

---

Output

Gap Report (coverage_gaps.json)

{
  "scanned_at": "2026-03-25T10:00:00Z",
  "project_path": "/home/dev/my_project",
  "total_source_functions": 42,
  "total_covered": 31,
  "gap_count": 11,
  "gaps": {
    "src/generator.py": [
      {"type": "function", "name": "render_skeleton"},
      {"type": "function", "name": "write_output_file"}
    ]
  }
}

Console Summary

Semantic Test Coverage Report
─────────────────────────────
Source functions found : 42
Already covered        : 31
Gaps identified        : 11
Skeleton files written : 3

New stubs written to:
  tests/test_generator.py      (2 functions)
  tests/test_analyzer.py       (7 functions)
  tests/test_main.py           (2 functions)

---

Source Reference

This plugin is derived from NathanMaine/semantic-test-coverage-agent, a CLI tool that scans Python projects via AST inspection and produces coverage gap reports and test skeletons.

---

License

MIT — see LICENSE.