unit-tests-skills
CLAUDE.md55 starsChanged 59 days agoMITView on GitHub
A collection of AI agent skills (not a runnable application) for generating high-quality unit tests. Skills are installed into target projects via npx openskills install or npx skills add. There is no build system, test suite, or application code here — only skill definitions and rule documents. The recommended sequence is generate-test-cases → generate-tests, chained by the agent: the first produces the plan, the second generates from that plan rather than analyzing the target again. It also puts the list…
In this repository3 other files it gives its agents.
- AGENTS.mdAGENTS.md
- generate-test-casesSkill
- generate-testsSkill
The file
We split this CLAUDE.md into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 93
Place general rules in rules/general/ (or rules/tests/general/ for generate-tests)
- Very specific: line 94
Place language-specific rules in rules/tests/{language}/unit/
- Very specific: line 96
When adding a new rule, also add it to the Rules Reference list in the relevant SKILL.md file(s)
- Distinctive: line 98
All changes require a PR with CODEOWNER approval.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
What This Repo Is
A collection of AI agent skills (not a runnable application) for generating high-quality unit tests. Skills are installed into target projects via npx openskills install or npx skills add. There is no build system, test suite, or application code here — only skill definitions and rule documents.
Repository Structure
skills/
generate-test-cases/ # Skill: analyze code → output test case list
SKILL.md # Skill definition (frontmatter + instructions)
rules/general/ # General testing rules
generate-tests/ # Skill: generate actual test code from cases
SKILL.md
rules/tests/
general/ # General testing rules (superset of generate-test-cases rules)
java/unit/ # Java-specific rules (JUnit 5, Mockito, AssertJ)
post-generation/ # Compilation verification rules
templates/
AGENTS-SNIPPET.md # Template users copy into their project's AGENTS.mdAvailable Skills
| Command | Purpose |
/generate-tests <target> | Generate unit tests for code. Runs the full workflow unattended: analyzes code, prints the test case list, generates test code, verifies it compiles and passes. Supports Java (JUnit 5, Mockito, AssertJ). |
/generate-test-cases <target> | Analyze code for test coverage and list needed test cases — without generating actual test code. Use for analysis-only. |
Workflow
The recommended sequence is generate-test-cases → generate-tests, chained by the
agent: the first produces the plan, the second generates from that plan rather than
analyzing the target again. It also puts the list in front of the user before any code
is written, so it can be reviewed. See AGENTS.md.
/generate-tests on its own is the fallback, and handles the complete workflow
internally:
- Analyzes code and outputs a structured test case list
- Generates test code
- Verifies the tests compile and pass
The run is unattended end to end: the test case list is printed, then the code follows, so the list stays auditable against the result.
/generate-test-cases is also useful on its own for analysis-only work (e.g., reviewing test coverage strategy without generating code).
Rules
Each skill's SKILL.md lists which rule files it reads. When a skill is invoked, the skill definition instructs the agent to read the rule files from the skill's own rules/ directory. The generate-tests skill has a superset of rules (includes Java-specific and post-generation rules).
Key rule topics:
- INCLUDE/EXCLUDE criteria (
test-case-generation-strategy.md) — what to test vs. skip - Naming (
naming-conventions.md) —{method}_{state}_{outcome}format - Structure (
general-principles.md) — Given-When-Then,actual/expectedprefixes - Existing test awareness (
existing-test-awareness.md) — check for existing tests before generating; match project conventions; avoid duplicates - Code context analysis (
code-context-analysis.md) — read DTOs, entities, enums, and other dependency classes before writing tests - Java specifics — JUnit 5 + Mockito + AssertJ;
@SpringBootTestis FORBIDDEN in unit tests; useArgumentCaptorto verify DTO/model fields; useany()only for irrelevant arguments;@WebMvcTestfor controllers (controller-test-rules.md) - Post-generation verification — compilation verification (
compilation-verification.md) AND test execution verification (test-execution-verification.md) — tests must both compile and pass
Plugin Distribution
This repository is also a Claude Code plugin and marketplace, declared in
.claude-plugin/. Users install it with:
/plugin marketplace add mavka-ai/unit-tests-skills
/plugin install unit-tests-skills@mavkaAny change to .claude-plugin/ or skills/ must be validated before pushing:
./scripts/validate-plugin.shCI enforces this on every PR touching those paths, and the community-marketplace
review pipeline runs the same claude plugin validate check — a failure here is
a rejected submission there.
Three non-obvious rules apply: validating the repo root does not validate
plugin.json, a mismatched entry name passes validation but breaks installs,
and version belongs in plugin.json only. See
AGENTS.md for the reasoning behind each — that
file is the single source of truth; do not duplicate its detail here.
Contributing
- Place general rules in
rules/general/(orrules/tests/general/for generate-tests) - Place language-specific rules in
rules/tests/{language}/unit/ - Keep general rules in sync: General rules exist in TWO locations (
generate-test-cases/rules/general/andgenerate-tests/rules/tests/general/). When adding or updating a general rule, copy the change to both directories - When adding a new rule, also add it to the Rules Reference list in the relevant
SKILL.mdfile(s) - After any change under a skill's
rules/, run./scripts/validate-rules.sh. It enforces the two points above — a reference with no file, a file with no reference, and general rules that have drifted apart. CI runs it on every PR. See AGENTS.md for what each check catches; that file is the single source of truth, do not duplicate its detail here - All changes require a PR with CODEOWNER approval; direct pushes to
mainare disabled
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

