Development skills

リポジトリにエージェント開発用の検証基盤を整える。`make check`、GitHub ActionsのCI、短いAGENTS.mdを、言語とツールを検出して作る。'harness-init'、'リポジトリを初期化して'、'検証コマンドを用意して'、新規リポジトリ作成直後や既存リポジトリへの導入時に使う。

The English summary is being prepared. Showing the original description.

Codex CLI（`codex exec`）に、モデル・reasoning effort・速度を指定して作業（調査、レビュー、相談、実装）を頼み、結果を受け取って要約する。「solで〜して」「astra xhighで調べて」「luna low fastで」「Codexで〜」「Codexに聞いて／やらせて」のように、呼び名（sol / astra / luna）やCodexを指定して作業を頼まれたときだけ使う。文中に普通の単語として「sol」「luna」が出てきただけのときは使わない。

The English summary is being prepared. Showing the original description.

Verify mini-program implementations and fixes with risk-calibrated evidence across static checks, unit tests, integration tests, state matrices, simulators, real devices, cloud environments, and release artifacts. Use when users ask to test, validate, accept, regression-check, quality-check, confirm readiness, or determine whether a mini-program feature is actually complete. Binds results to a source and build fingerprint, records commands and observable evidence, separates passed, failed, blocked, and not-run layers, prioritizes the next highest-information check, and never converts local success into device, cloud, release, or formal acceptance claims.

The English summary is being prepared. Showing the original description.

Orchestrate evidence-first engineering for WeChat and other mini programs from a vague idea or existing repository through project discovery, product specification, architecture, implementation, debugging, device adaptation, verification, and release readiness. Use when users ask to build a mini program from zero to one, take over an existing mini-program project, deliver a cross-stage feature, diagnose project status, coordinate development and testing, or judge whether a mini program is ready to upload or release. Enforces fact discovery, change boundaries, stage gates, evidence-calibrated status, sensitive-information redaction, and continuity after side questions.

The English summary is being prepared. Showing the original description.

Implement bounded features and fixes in existing or new mini-program codebases while preserving repository rules, user-owned changes, approved behavior, assets, framework conventions, and evidence boundaries. Use when users ask to write or modify mini-program code, implement a confirmed specification or architecture, add a scoped feature, remove behavior semantically, update internal documentation required by a change, or carry out a well-defined small fix that does not require root-cause investigation. Establishes a baseline and change boundary, applies test-driven small steps, distinguishes source from generators and build artifacts, verifies the affected contract, and never reports source completion as device validation, formal acceptance, upload, or release.

The English summary is being prepared. Showing the original description.

Govern mini-program release readiness and evidence across source branches, semantic versions, build artifacts, test flags, debug paths, sensitive data, permissions, privacy declarations, platform uploads, review submissions, staged versions, production releases, and rollback plans. Use when users ask for a release checklist, packaging, export, version bump, upload readiness, submission readiness, launch status, rollback preparation, or a precise judgment of what has and has not shipped. Defaults to read-only preflight, requires separate explicit authorization for every external state change, and never treats a Git push, successful build, platform upload, review submission, approval, or production publication as interchangeable.

The English summary is being prepared. Showing the original description.

Convert a vague mini-program idea, feature request, or partially documented product into an evidence-calibrated product specification with target users, core problem, version-one scope, main and exception flows, page responsibilities, state matrices, and testable acceptance criteria. Use when users ask to define an MVP, clarify requirements, organize product flows, specify page behavior, resolve ambiguous product states, or prepare a stable handoff before architecture or implementation. Separates facts, user decisions, assumptions, unknowns, and future ideas; never invents product logic merely to complete a screen or technical plan.

The English summary is being prepared. Showing the original description.

Analyze WeChat Mini Program platform constraints and evidence layers for an idea or existing repository, including source configuration, build output, WeChat DevTools loading, device behavior, permissions, privacy declarations, cloud or backend linkage, upload, trial, review, and production status. Use when users ask about WeChat-specific configuration, project roots, compile paths, capability permissions, privacy compliance, official platform rules, upload or release readiness, or why source and platform behavior differ. Performs read-only evidence classification by default, verifies time-sensitive or high-risk claims against current official WeChat sources, redacts identifiers and secrets, and never upgrades one evidence layer into another.

The English summary is being prepared. Showing the original description.

Translate a confirmed mini-program product specification or bounded feature into an implementable architecture covering modules, pages, components, services, canonical state sources, data models, interfaces, permissions, failures, caching, concurrency, idempotency, migration, rollback, external dependencies, and architecture decision records. Use when users ask how to structure a mini program, design data or APIs, divide frontend and cloud responsibilities, evolve an existing system safely, or prepare implementation after product semantics are stable. Preserves product meaning, current repository constraints, user-owned changes, and verifiable acceptance behavior rather than redesigning the product for technical convenience.

The English summary is being prepared. Showing the original description.

Apply evidence-first engineering discipline to any agent-built software project: fact discovery before action, evidence-calibrated status reporting, explicit change boundaries and stage gates, controlled confirmation for risky or external actions, sensitive-information redaction for anything shared publicly, and resumable continuity after interruptions. Use when users ask an agent to take over an existing codebase, deliver a feature across stages, verify whether work is actually complete, judge release readiness, or keep conclusions honest about what is proven versus assumed. This foundation skill is domain-neutral; vertical suites (for example mini-program engineering) build on it by adding domain facts and platform rules.

The English summary is being prepared. Showing the original description.

Perform a read-only, evidence-first intake of an existing mini-program repository before planning or modifying code. Use when users ask to take over, understand, audit, resume, scope, or continue a WeChat or other mini-program project; when project facts may conflict with historical documents; or when an agent needs the framework, rules, Git state, build path, risks, unknowns, protected behavior, and change boundary. Produces a project fact map and handoff without changing code, installing dependencies, building artifacts, or claiming runtime, device, cloud, upload, acceptance, or release status.

The English summary is being prepared. Showing the original description.

Diagnose and fix mini-program failures through reproducible, evidence-driven root-cause analysis across source, state, asynchronous timing, caches, mock data, build versions, platform behavior, devices, cloud calls, and external services. Use when users report white screens, crashes, stale or jumping values, missing data, intermittent behavior, permission failures, performance stalls, device-only defects, API or cloud-function errors, source-versus-build mismatches, or regressions whose cause is not yet established. Freezes the observation environment, ranks competing hypotheses, runs discriminating checks, creates a failing regression test before repair, covers analogous states, and never hides a known failure with longer waits, swallowed errors, forced success, or cosmetic patches.

The English summary is being prepared. Showing the original description.

Use this when managing Python dependencies with uv, running tests with pytest, linting with ruff, formatting code, type checking with mypy, or executing notebooks.

The English summary is being prepared. Showing the original description.

Use this when creating, editing, or reviewing Python code — including type hints, docstrings, naming conventions, imports, error handling, and code structure.

The English summary is being prepared. Showing the original description.

Trigger.dev client/frontend surface: subscribe to runs in realtime (runs.subscribeToRun and the @trigger.dev/react-hooks hook useRealtimeRun), consume metadata and AI/text streams in React (useRealtimeStream), trigger tasks from the browser (useTaskTrigger, useRealtimeTaskTrigger), and mint scoped frontend credentials with auth.createPublicToken / auth.createTriggerPublicToken. Load when wiring a frontend (React/Next.js/Remix) or backend-for-frontend to show live run progress, status badges, token streams, trigger buttons, or wait-token approval UIs. NOT for writing the backend task itself (streams.define / metadata.set is trigger-authoring-tasks territory); this is the consumer side.

The English summary is being prepared. Showing the original description.

Clerk Backend REST API explorer and executor. Browse tags, inspect endpoint schemas, and execute authenticated requests. Use when listing users, managing organizations, or calling any Clerk API endpoint.

The English summary is being prepared. Showing the original description.

Operate the Clerk CLI (`clerk` binary) for authentication, user/org/session management, deploy verification, instance config, env keys, and any Clerk Backend or Platform API call. Use when the user mentions Clerk management tasks, "list clerk users", "create a clerk user", "update organization", "pull clerk config", "clerk env pull", "clerk doctor", "clerk deploy", "clerk deploy status", "clerk api", or any ad-hoc Clerk API request. Prefer the CLI over raw HTTP: it handles auth, key resolution, app/instance targeting, and formatting automatically.

The English summary is being prepared. Showing the original description.

Guides and best practices for working with Neon Serverless Postgres. Covers setup, connection methods, branching, autoscaling, scale-to-zero, read replicas, connection pooling, Neon Auth, and the Neon CLI, MCP server, REST API, TypeScript SDK, and Python SDK. Use when users ask about "Neon setup", "connect to Neon", "Neon project", "DATABASE_URL", "serverless Postgres", "Neon CLI", "neon", "Neon MCP", "Neon Auth", "@neondatabase/serverless", "@neondatabase/neon-js", "scale to zero", "Neon autoscaling", "Neon read replica", or "Neon connection pooling".

The English summary is being prepared. Showing the original description.

Advanced Next.js patterns - middleware, Server Actions, caching with Clerk.

The English summary is being prepared. Showing the original description.

Review a Python solution against a markdown spec, run it, and report bugs.

The English summary is being prepared. Showing the original description.

Create isolated git worktree for parallel development in monorepos.

The English summary is being prepared. Showing the original description.

Create or update Codex skills optimized for Skillmark benchmarks. Use for new skills, skill scripts, references, benchmark optimization, extending Codex's capabilities.

The English summary is being prepared. Showing the original description.

Build with TanStack Start (full-stack React framework), TanStack Form (headless form management), and TanStack AI (AI streaming/chat). Use when creating TanStack projects, routes, server functions, forms, validation, or AI chat features.

The English summary is being prepared. Showing the original description.

Utilize MCP server tools with intelligent discovery and execution.

The English summary is being prepared. Showing the original description.