starshine-mb

Implement, port, or sign off Starshine optimizer passes with the repo-standard TDD, registry, Binaryen semantic parity, performance, validity, and documentation workflow. Use when working on pass creation, pass porting, pass parity, pass fuzz compare failures, optimizer registry wiring, preset changes, or pass signoff.

Commit current work using Starshine's commit policy. Use when the user asks to commit, create commits, save changes to git history, or prepare a commit from the current worktree.

Create a continuation thread with default thread parameters and a comprehensive handoff of the current task. Use when the user says "/handoff", asks to hand off the current work, continue in another thread, or wants a fresh thread with full context.

Ultra-compressed communication mode. Cuts token usage ~75% by dropping filler, articles, and pleasantries while keeping full technical accuracy. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman.

Generate multiple radically different interface designs for a module using parallel sub-agents. Use when user wants to design an API, explore interface options, compare module shapes, or mentions "design it twice".

Grilling session that challenges your plan against the existing domain model, sharpens terminology, and updates documentation (CONTEXT.md, ADRs) inline as decisions crystallise. Use when user wants to stress-test a plan against their project's language and documented decisions.

Edit and improve articles by restructuring sections, improving clarity, and tightening prose. Use when user wants to edit, revise, or improve an article draft.

Set up Claude Code hooks to block dangerous git commands (push, reset --hard, clean, branch -D, etc.) before they execute. Use when user wants to prevent destructive git operations, add git safety hooks, or block git push/reset in Claude Code.

Triage GitHub issues through a label-based state machine with interactive grilling sessions. Use when user wants to triage issues, review incoming bugs or feature requests, prepare issues for an AFK agent, or manage issue workflow.

Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".

Explore a codebase to find opportunities for architectural improvement, focusing on making the codebase more testable by deepening shallow modules. Use when user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more AI-navigable.

Migrate existing type-test suites to ShoeHORN-style assertions and fixtures. Use when user wants to move from older TypeScript type-test tools to ShoeHORN, replace brittle type assertions, or standardize type-level tests.

Use an Obsidian vault as the working memory for a project by searching notes, surfacing context, and creating or updating notes as durable memory. Use when the user wants work grounded in an Obsidian vault.

Perform a focused QA pass on recent changes by looking for regressions, edge cases, and user-visible rough edges. Use when user asks for QA, a review pass, or bug-hunting after implementation.

Turn a rough desire to refactor into a concrete, reviewable plan with goals, risks, sequencing, and acceptance criteria. Use when user wants help planning a refactor before coding.

Turn existing code or docs into exercises with clear starter state, prompts, and expected outcomes. Use when user wants workshop material, learning exercises, or kata scaffolding.

Set up pre-commit checks so formatting, linting, or targeted validation runs before commits. Use when user wants safer commits or wants to enforce checks locally.

Practice strict test-driven development with tiny loops, clear interfaces, and refactoring only after green tests. Use when user asks for TDD or wants work done test-first.

Turn a document, plan, or backlog into concrete GitHub issues with clear scope and acceptance criteria. Use when user wants to break work into issues.

Turn a rough product idea or feature request into a concise PRD with goals, user needs, constraints, and success criteria. Use when user wants a PRD.

Investigate a single issue thoroughly, clarify reproduction and scope, and recommend the right next step. Use when user wants help triaging one issue in depth.

Discover, sharpen, and standardize the project's terminology so code, docs, and discussion use the same words for the same concepts. Use when user wants a glossary or more precise language.

Create a new skill by defining when it should trigger, what workflow it follows, and any supporting reference files it needs. Use when user wants to author a new Pi skill.

Step back from the current implementation detail, summarize the broader situation, and recommend the next high-leverage direction. Use when user is stuck in details or asks to zoom out.