菜单
Define reliable AI provider harnesses with mock providers, schemas, retries, invalid JSON recovery, persistence tests, and provider contracts. Use for LLM, image, voice, subtitle, render, and pipeline provider work.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Define UI design specs, route contracts, visual states, accessibility expectations, and UI/API synchronization before frontend implementation. Use for new screens, redesigns, visual QA, or UI contract updates.
Automatically synchronize documentation with code changes across all project phases. Detects file changes, categorizes update types, auto-generates API docs, changelog entries, implementation handoffs, and validates cross-reference integrity. Use after any phase implementation.
Initialize and operate a project planning harness for Codex. Use this skill when the user types /init, asks to create a new project, add a feature, fix a bug, plan work, generate tests first, update docs, track phases, review changes, audit a repository, or manage architecture decisions.
Create premium frontend web designs and usable first-screen experiences for new websites, web apps, dashboards, tools, and landing pages. Use when Codex is asked to design or build a new frontend UI, create a fresh page or app experience, choose a visual direction for a new product surface, or turn a product brief into implementation-ready React, Next.js, Tailwind, CSS, or HTML.
Create decision-complete plans for Codex harness work, including tests, fixtures, contracts, memory updates, verification, and recovery. Use for new features, refactors, bug fixes, audits, or multi-phase autonomous work.
Prepare Codex harness package releases with version checks, package dry-runs, npm publish readiness, release notes, verification evidence, and rollback notes. Use before publishing or tagging releases.
| name | genesis-ai-provider |
| description | Define reliable AI provider harnesses with mock providers, schemas, retries, invalid JSON recovery, persistence tests, and provider contracts. Use for LLM, image, voice, subtitle, render, and pipeline provider work. |
Make AI provider behavior deterministic under tests and recoverable under failures.
Use before adding or changing AI providers, provider adapters, retry logic, schema parsing, or persistence.
Do not use for purely static documentation changes.
Provider name, request shape, expected response, error modes, retry policy, and persistence target.
Provider contract, mock provider fixture, schema validation, retry test, invalid JSON recovery test, and persistence test.
Mock success, provider error, retry, invalid JSON recovery, and persistence tests.
Create provider fixtures under fixtures/agents, fixtures/images, fixtures/tts, or related directories.
Update contracts/agents/ or provider-specific contracts.
Update providers summary, pipeline flow, and test matrix.
Read provider summary and contract before provider code.
Provider output is schema-validated and failure recovery is tested.
Trusting raw model JSON, skipping retry tests, and using live provider calls in normal tests.
If provider output is invalid, capture fixture, add parser recovery test, then patch parser or contract.
