MU

mutation-testing

Automates mutation testing, test generation, and bug tracking for Clojure projects.

Install

mkdir -p .claude/skills/mutation-testing && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/2604" && unzip -o skill.zip -d .claude/skills/mutation-testing && rm skill.zip

Installs to .claude/skills/mutation-testing

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Run mutation testing on a Clojure namespace, generate tests to kill surviving mutations, and open draft PRs with Linear issue tracking.
135 charsno explicit “when” trigger
Advanced

Key capabilities

  • Executes Clojure-specific mutation testing
  • Identifies surviving mutations in source code
  • Automates test generation to cover missed mutations
  • Links test improvements to Linear issues

How it works

It runs local Clojure mutation engine scripts to identify code gaps and then invokes an LLM to generate tests that invalidate the surviving mutants.

Inputs & outputs

You give it
Target Clojure namespace
You get back
Mutation report and draft PR link

When to use mutation-testing

  • Identify weak test coverage
  • Automate mutation kill-rate improvements
  • Create draft PRs for test fixes

About this skill

Mutation Testing Skill

This skill runs mutation testing end-to-end: generates a coverage report, groups functions, shells out to claude -p to write tests, verifies mutations are killed, and creates draft PRs with Linear issues.

Prerequisites

  • A running nREPL connected to the Metabase dev environment
  • LINEAR_API_KEY environment variable set with a valid Linear personal API key
  • gh CLI authenticated with GitHub
  • claude CLI available on PATH

Reference Files

  • dev/src/dev/coverage.clj — mutation testing engine (generates reports, runs mutations)
  • dev/src/dev/mutation_testing.clj — orchestration, Linear API, PR helpers

Invocation

The argument is a Clojure namespace, optionally followed by a base branch:

/mutation-testing metabase.lib.order-by
/mutation-testing metabase.lib.order-by --base-branch release-x.52.x
/mutation-testing metabase.lib.order-by --project-id abc-123
/mutation-testing metabase.lib.order-by --base-branch release-x.52.x --project-id abc-123
  • --base-branch: defaults to the value in config (set via set-config!), or "master" if not configured.
  • --project-id: if provided, issues are added to this existing Linear project instead of creating a new one.

Steps

1. Parse arguments

Parse $ARGUMENTS to extract:

  • namespace (required) — the first positional argument
  • --base-branch (optional) — if present, the next argument is the base branch name
  • --project-id (optional) — if present, the next argument is an existing Linear project ID

2. Load and configure

(require '[dev.mutation-testing :as mut-test] :reload)
(require '[dev.coverage :as cov] :reload)

If this is the first invocation (or the REPL was restarted), set up the Linear team:

(mut-test/list-teams!)
(mut-test/set-config! {:team-id "<id>"})

3. Run

(mut-test/run! '<target-ns>)
;; Or with options:
(mut-test/run! '<target-ns> {:base-branch "release-x.52.x"})
(mut-test/run! '<target-ns> {:project-id "abc-123"})
(mut-test/run! '<target-ns> {:base-branch "release-x.52.x" :project-id "abc-123"})

Pass the opts map with :base-branch and/or :project-id if those flags were provided in the arguments.

This will:

  1. Generate a baseline mutation testing report
  2. Create a Linear project for the namespace
  3. Group functions by coverage relationships
  4. For each group: create branch → invoke Claude to write tests → verify → retry if needed → commit & push → create Linear issue → create draft PR
  5. Print a summary with PR links

4. Handle failures

If a group fails mid-processing, run! catches the error and continues to the next group. To retry a single group manually:

;; Get the parsed namespace info
(def parsed (mut-test/parse-namespace '<target-ns>))

;; Run coverage to get the data
(def coverage-results (cov/test-namespace (:target-ns parsed) [(:test-ns parsed)]))

;; Group and find the one you want
(def groups (mut-test/group-functions coverage-results))

;; Process just that group
(mut-test/process-group! parsed (nth groups <index>))

5. Verify the prompt (dry run)

To inspect what Claude will see without invoking it:

(def parsed (mut-test/parse-namespace '<target-ns>))
(def coverage-results (cov/test-namespace (:target-ns parsed) [(:test-ns parsed)]))
(def groups (mut-test/group-functions coverage-results))
(println (mut-test/build-test-prompt
           (merge (select-keys parsed [:target-ns :test-ns :source-path :test-path])
                  (select-keys (first groups) [:fn-names :mutations]))))

Configuration

  • Team ID: Set once per REPL session via (mut-test/list-teams!) and (mut-test/set-config! {:team-id "..."})
  • Project ID: Set automatically when run! calls create-project-for-namespace!
  • Base Branch: Defaults to "master". Override via (mut-test/set-config! {:base-branch "release-x.52.x"}) or pass {:base-branch "..."} as the second arg to run!
  • Project ID: If set (via set-config! or run! opts), run! reuses the existing Linear project instead of creating a new one

When not to use it

  • Projects using non-Clojure languages
  • Environments without nREPL access

Prerequisites

Clojure runtimeLINEAR_API_KEYGh CLI

Limitations

  • Requires specific project structure defined in Metabase dev environment
  • Execution time scales linearly with test suite size

How it compares

It creates a tight feedback loop between mutation detection, code generation, and project management tools like Linear.

Compared to similar skills

mutation-testing side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
mutation-testing (this skill)55moReviewAdvanced
verification-before-completion99moNo flagsBeginner
workflow-patterns12moNo flagsIntermediate
tdd-workflows-tdd-refactor14moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry