v3/skills/v3-graveyard-verify/SKILL.md
# Skill: /v3-graveyard-verify **Purpose**: Post-generation verification that cross-references the generated test code against GRAVEYARD.md to catch known mistakes before runtime ## Input - **Required**: Path to the generated test file ## Output - **Modified File**: Test file edited in-place with fixes for any GRAVEYARD violations found - **Report**: Summary of violations found and fixes applied (conversational output) ## Role You are a **code reviewer specializing in catching known mistakes*
npx skillsauth add ruclo/thesis v3/skills/v3-graveyard-verifyInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Purpose: Post-generation verification that cross-references the generated test code against GRAVEYARD.md to catch known mistakes before runtime
You are a code reviewer specializing in catching known mistakes. You have access to GRAVEYARD.md, which documents every mistake made during previous test generation runs. Your job is to read the generated code and flag any pattern that matches a documented mistake, then fix it before the code ever reaches pyright or runtime execution.
The code generation phase (Phase 3) is instructed to apply GRAVEYARD lessons, but it operates from memory of the exploration phase and may miss specific patterns. This dedicated verification phase provides a second pass with direct, focused comparison between the generated code and every GRAVEYARD entry. It catches mistakes that slip through generation, reducing the number of pyright and runtime iterations needed.
Read GRAVEYARD.md from the repository root. For each entry, extract:
If GRAVEYARD.md does not exist, report that there are no past mistakes to verify against and exit successfully.
Read the entire generated test file.
For each GRAVEYARD entry, check whether the generated code contains the "Wrong Code" pattern or violates the "How to Avoid" rule:
FOR each graveyard_entry in GRAVEYARD.md:
# Check for exact pattern match
IF generated_code contains graveyard_entry.wrong_code:
FLAG as violation
# Check for semantic rule violation
IF generated_code violates graveyard_entry.how_to_avoid:
FLAG as violation
For each violation found:
Report how many entries were checked, how many violations were found, and what fixes were applied.
/v3-graveyard-verify <test_file_path>
High - Can verify any generated test file against GRAVEYARD.md
development
# Skill: /v3-test-heal **Purpose**: Self-healing loop that runs the generated test against a real cluster, analyzes failures from logs, consults documentation and repository code to fix issues, and records lessons learned in GRAVEYARD.md ## Input - **Required**: Path to the generated test file (e.g., `tests/virt/cluster/test_vnc_screenshot.py`) - **Optional**: `--max-iterations N` - Maximum heal attempts (default: 3) ## Output - **Modified File**: Test file edited in-place with fixes applied
development
# Skill: /v3-pyright-heal **Purpose**: Universal Python type checker and self-healing fixer for ANY Python file ## Input - **Required**: Path to Python file (`.py` or `.ipynb`) - **Optional**: `--max-iterations N` - Maximum fix attempts (default: 10) - **Optional**: `--strict` - Use strict type checking mode ## Output - **Modified File**: Same file, edited in-place - **Exit Code**: 0 if clean, 1 if max iterations reached with errors - **Log**: Summary of fixes applied ## Implementation ###
development
# Skill: /v3-generate-std > _This document was created with the assistance of Claude (Anthropic)._ You are a Senior QE Engineer responsible for creating Software Test Descriptions (STDs) for OpenShift Virtualization. Given a high-level Software Test Plan (STP), generate test code stubs with comprehensive docstrings that serve as the STD. **Important:** The generated test code should follow the structure and patterns from the [openshift-virtualization-tests](https://github.com/RedHatQE/openshi
development
# Skill: /v3-generate-pytest **Purpose**: Generate executable pytest code from Software Test Description (STD) using repository context ## Input - **Required**: Path to STD markdown file - **Prerequisite**: Repository context should be explored using `/v3-explore-test-context` first - **Optional**: `--output-dir` - target directory (default: infer from repository conventions) ## Output - **File**: `test_<feature_name>.py` - **Location**: Determined by repository conventions (e.g., `tests/<dom