claude-code-plugins

Run the complete Contree arc from idea to independently reviewed working software without routine operator phase transitions: change sets the contract, sync audits it, tdd closes every gap, mutation and completion sync prove the result, and second-opinion findings are repaired before completion. TRIGGER when: the user wants the full workflow, asks to take an idea through implementation, asks to make the project complete, or wants change/sync/tdd handled end to end.

Set up fast test feedback for coding agents by configuring normal, journey, and impacted-test commands plus project save hooks. TRIGGER when: an operator asks to configure a test framework, test commands, test output, changed-test selection, test hooks, or Docker-backed tests.

Comprehensively install Contree's coding-agent feedback loops through a dynamic, operator-guided workflow. TRIGGER when: the operator asks for full, complete, comprehensive, or luxury Contree setup, or starts a project and wants all steering configured.

Set up fast conventional code linting, automatic repair, CI enforcement, and synchronous coding-agent feedback. TRIGGER when: an operator asks to add, configure, improve, or fix linting or code-quality feedback.

Set up test trees and durable project-local coding-agent steering without implementing tests. TRIGGER when: an operator asks to establish, discover, backfill, configure, or set up test trees for an existing or new project.

Set up and prove fast mutation-test feedback, then strengthen Domain and Use-case tests until the agreed threshold passes. TRIGGER when: an operator asks to configure, run, optimize, or fix mutation testing.

Identify and resolve drift between operator intention, test trees, tests, implementation, and the mental model. TRIGGER when: the user asks about drift, gaps, staleness, completeness, whether trees match code, whether the project is in sync, what is missing, or what should be removed.

Generate a single image that explains the current change to a human, using OpenAI's gpt-image-2 model — foregrounding the technical substance the change touches (contracts, databases, behaviour, test trees). TRIGGER when: the user runs /diff-for-humans, or asks to see, picture, illustrate, or visualise the current change or diff.

Set expected behaviour by writing or modifying test trees before code changes. TRIGGER when: the user describes a feature, capability, behaviour change, removal, or modification they want — even loosely (e.g. 'I want X', 'let's add Y', 'can we make it do Z', 'change how X works', 'remove this behaviour').

Implement behaviour from test trees through outside-in, consumer-driven RED-GREEN-REFACTOR. TRIGGER when: implementing behaviour, writing code, writing tests, fixing a failing test, or closing an implementation gap from existing test trees.

Get an independent review of completed work from OpenAI's gpt-5.6-sol with high reasoning effort — sends the work and test-tree contract through the Responses API with OPENAI_API_KEY and surfaces its critique. TRIGGER when: sync has just finished, the user asks for a second opinion, an independent review, or a sanity check on completed work before a PR or release.

Run and evaluate the contree plugin's Docker-based journey tests. TRIGGER when the user mentions 'journey test', 'journey tests', 'docker-run', 'test-kinds-workflow', 'test transcript', 'verify against trees', 'review the transcript', asks to 'kick off / run the end-to-end test', asks whether the plugin behaves correctly end-to-end, or asks to evaluate a completed test run.

Compose mental-model and test-tree setup, then implement every existing-project tree through a second subagent wave. TRIGGER when: the operator asks to bootstrap, backfill, discover, or implement a project's existing behaviour as test trees and tests.

Discover and reconcile an existing project's mental model, or create its empty seven-section home, then install durable project-local steering hooks. TRIGGER when: an operator asks to set up, discover, bootstrap, or enforce a project mental model.

Set up executable hexagonal-boundary linting, CI enforcement, and coding-agent Stop feedback. TRIGGER when: an operator asks to add, configure, improve, or verify architecture linting or dependency boundaries.

Repair every reported architecture violation with partitioned subagent work and rerun boundary feedback until green. TRIGGER when: architecture lint reports violations or an operator asks to fix dependency direction, purity, composition-root, or cycle problems.

Publish a plugin from this repo end-to-end: review commits since the last tag, draft release notes, run the publish script. TRIGGER when the user says 'publish <plugin>', 'release <plugin>', 'cut a release', 'ship contree/trunk-sync', or asks to publish without specifying how.