Skills Plugins MCP Prompt Model 博客 我的中心
开发编程 #design #ai

using-bridge

Use when any Bridge command is invoked (make, fix, done, setup, drop, status) or any Figma / design-system / compiler / Bridge workflow topic is raised. Sets command priorities and iron laws (compiler-only, semantic tokens only, verification-before-ship).

DeepseekModel 官方收录技能 质量 优秀 · 78 v1.0.0

获取

https://deepseekmodel.com/api/download.php?id=noemuch-bridge-skills-using-bridge-skill-md&format=skill
下载 .skill 标准格式,含 system_prompt 与 model_config,导入任意 Agent 框架即可使用
.skill 文件中 system_prompt 字段的实际内容。
name using-bridge description Use when any Bridge command is invoked (make, fix, done, setup, drop, status) or any Figma / design-system / compiler / Bridge workflow topic is raised. Sets command priorities and iron laws (compiler-only, semantic tokens only, verification-before-ship). Using Bridge Bridge is a compiler-driven design workflow for generating Figma designs and maintaining a design system via Claude Code. The compiler (at lib/compiler/compile.ts ) enforces all 26 Figma Plugin API rules, so Claude NEVER writes raw Plugin API code and NEVER hardcodes primitive values. This skill is force-loaded at every SessionStart via hooks/session-start . Its job is to establish the discipline before any action skill runs. It is deliberately small (~400 tokens) to keep the fixed per-session cost low. Command Map User intent (keywords) Route to "make", "design", "create", "build", "generate", "new component", "new screen" generating-figma-design "fix", "correct", "learn", "diff", "what changed", "I adjusted" learning-from-corrections "done", "ship", "ship it", "finish", "complete" shipping-and-archiving "setup", "setup bridge", "extract", "extract DS", "onboard", "initialize", "bootstrap" extracting-design-system "drop", "abandon", "cancel" inline Drop Procedure (this skill) "status", "what's next", "workflow" inline status logic (this skill) Drop Procedure (inline) drop is handled inline here — it is small enough not to warrant its own skill. Invoke when the user says "drop", "abandon", or "cancel". Confirm. Ask: "Sure you want to drop {name}?" Capture learnings. If a snapshot exists, offer to run fix first to capture corrections before archiving. Document drop reason. Append a drop: block to the CSpec with date , reason , and learnings . Archive. Move specs/active/{name}.cspec.yaml → specs/dropped/{name}.cspec.yaml . Move the snapshot JSON too if it exists. Update history. Append {ISO date} | {name} | DROPPED | {reason} to specs/history.log . Cleanup. Remove /tmp/bridge-scene-{name}.json if present. Output template: ## Dropped: {name} Reason: {reason} CSpec archived: specs/dropped/{name}.cspec.yaml Learnings: {captured | skipped} Ready for the next design. Run: `make <description>`. Skill Priority Process first, then action. For exploratory or ambiguous requests, brainstorm the intent first before implementing. For a clear directive that maps to a command in the table above, route directly. Verification before completion. No "done" without evidence (see Iron Laws below). Minimal context. Load only the references needed for the current action. See each action skill's ## Verification section. Iron Laws (non-negotiable) NEVER write raw Figma Plugin API code. All scene graph JSON must pass through `lib/compiler/compile.ts`. Violations require explicit human approval before any execution to Figma. NEVER use hardcoded primitive values. Only semantic DS tokens (`$color/...`, `$spacing/...`, `$text/...`, `$comp/...`). The compiler emits `RESOLVE_TOKEN_NOT_FOUND` for any unresolved reference. NEVER claim "done" without: (a) compiler exit 0, (b) screenshot taken in this turn, (c) user confirmation of visual correctness. "Looks right" / "should pass" / "I'm confident" are forbidden — show the evidence. NEVER read `figma-api-rules.md`. The compiler enforces all 26 rules. This file does not exist in v6. NEVER reuse a Figma `nodeId` from a previous session. Node IDs are session-scoped — re-search. Red Flags — Rationalization → Reality Rationalization Reality "I'll just hardcode this hex once" Always use a semantic token. No exceptions. "The compiler is overkill for this tiny thing" The compiler is the only path. "Skip the screenshot, it's obviously right" 'Looks right' ≠ 'is right'. "I remember this nodeId from my last session" Node IDs are session-scoped. Re-search. "I'll use figma-api-rules.md for context" That file is forbidden. Compiler owns all rules. "The user approved, I can skip the compile exit code check" Compile exit 0 is Gate A. Independent of user approval. "Let me write a small inline Plugin API script for this fix" No inline scripts. Scene graph → compiler → execute. References Compiler reference: references/compiler-reference.md (repo-root) Transport adapter: references/transport-adapter.md (repo-root) Verification gates: references/verification-gates.md (repo-root) Red Flags catalog: references/red-flags-catalog.md (repo-root) Conversation Language Rule Conversation with the user: their language (detect from context). All generated artifacts (KB files, CSpecs, guides, learnings, recipes, scene graphs, docs, specs, plans): English only . This rule is non-negotiable per Bridge's artifact policy.
Agent 识别该技能的关键词,点击任意一个即可复制。

该技能未提供触发词。

下载的 .skill 包内含以下字段。
字段 说明
format格式标识(skill/v1)
skill_id技能唯一 ID
name技能名称
version版本号
description技能描述
category所属分类(数组)
trigger_words触发词列表
tags标签列表
source来源标识
source_url来源链接(本页地址)
exported_at导出时间(每次下载生成)
system_prompt系统提示词正文
model_config模型参数:provider / model / temperature / max_tokens / top_p
examples示例
install_guide各平台导入说明(Coze / Dify / Claude / 自定义框架)
同一份技能可按不同平台格式导出。
.skill 标准格式,含 system_prompt 与 model_config,导入任意 Agent 框架即可使用 下载
.skillpro 增强格式,额外含脚本 / 工具 / 依赖 / 钩子占位 下载
.json 纯 JSON 导出,只含 system_prompt 与模型参数 下载
Coze 带 frontmatter 的 Markdown,Coze 平台导入用 下载
Dify Dify DSL,创建应用后直接导入 下载

每日精选 Skill 推荐,免费送到你邮箱

输入邮箱,每天接收一个精选 AI Agent 技能推荐。完全免费,持续更新。

验证码 --

提交后我们会发送一封确认邮件,点击邮件里的链接才会开始收信。

完全免费,取消任意时间。我们不会发送垃圾邮件。