pincushion-plugin

MCP设计创作

提供方 jcooley8 · 包名 pincushion-mcp

设计创作 stdio · 本地进程 API Key 社区 / 第三方 Claude DesktopClaudeCursorVS Code
查看接入方式

Pincushion MCP Server is an MCP server that turns visual page pins into AI agent work packets, allowing AI coding agents in Claude Code, Cursor, VS Code, Windsurf, or any MCP client to read, fix, and verify issues directly on live app pages for design feedback, code fixes, accessibility review, and deployment verification.

一句话选型结论

按资料字段整理

pincushion-plugin 属于设计创作类 MCP 服务器,stdio · 本地进程,API Key,可直接接入 Claude Desktop、Claude、Cursor 等,有可直接执行的安装命令。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类设计创作 传输方式stdio · 本地进程 鉴权API Key 部署方式本地部署
GitHub 星标 1 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

设计创作
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

Local Deployment MCP Server AI Coding Agent JavaScript Visual Feedback Code Fixing Development Collaboration Deployment Verification Design Review Accessibility Review

支持客户端

适用场景

Local DeploymentMCP ServerAI Coding Agent

安装与配置

stdio · 本地进程 API Key
客户端在本机拉起子进程,数据不出本机
npx pincushion-mcp setup

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    stdio 服务在本地起进程,需要先装好 node,再按安装命令拉取依赖。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
出图与素材

生成配图、图标与素材。

设计稿处理

读取设计稿、批量改尺寸或导出。

内容生成

按主题生成文案与视觉草稿。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Claude Desktop、Claude、Cursor,换客户端不用重新研究配置。

先想清楚的地方

  • stdio 方式要求本地装好运行时(如 Node.js),换机器要重新装一遍。
  • 使用前要准备好API Key对应的凭据,并按最小权限配置。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在设计创作场景里让模型直接调用外部能力的人。
  • 已经在用 Claude Desktop、Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 完全不想在本地装环境的人。
  • 不方便管理密钥或第三方授权的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
pincushion-plugin当前 未标注 stdio · 本地进程 本地部署 有 2026-10-08
Cursor Talk To Figma MCP Da1c7911 40 未标注 未标注 无 2026-05-17
Blender MCP 17 stdio · 本地进程 本地部署 有 2026-05-18
Nova Canvas MCP Server 2 stdio · 本地进程 本地部署 有 2026-05-17
Figma Context MCP 3aaa8d28 未标注 未标注 支持远程部署 无 2026-05-17
kreuzberg 未标注 stdio · 本地进程 本地部署 有 2026-04-12

服务器档案

设计创作
字段来自项目仓库与公开索引,中文说明由本站标注
提供方jcooley8上游标注的维护方 jcooley8
分类设计创作本站按上游分类展示
连接方式stdio · 本地进程客户端在本机拉起子进程,数据不出本机
鉴权方式API Key远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时pincushion-mcp · node本地安装时用到的包标识与运行时
部署类型本地部署上游给出的部署形态
最近核验2026-10-08上游最后核验时间

详细介绍

15.9 KB
来自项目仓库与上游目录的原文,未做删改

Pincushion MCP Server

The implementation-context layer for AI-native development. Stakeholders drop visual pins on any page of your live app; your AI coding agent reads each pin through MCP and ships the fix — in Claude Code, Cursor, VS Code, Windsurf, or any MCP client.

What makes a Pincushion pin different

A pin isn't a feedback item — it's an agent work packet. Each one carries everything an agent needs to implement the change without a back-and-forth:

  • URL + element selector — exactly what, exactly where
  • Screenshot + viewport + DOM snippet — the visual and structural context
  • Thread + project context — the conversation and the codebase it lives in
  • Likely files + acceptance criteria — where to look, and how to know it's done

The loop closes itself: a stakeholder pins it → your agent reads it via MCP and fixes it in your IDE → the resolve records the commit, branch, and PR → an optional post-deploy critique verifies the fix actually landed.

This server is also how Pincushion AI runs design/copy/a11y critiques on a live page and writes the pins straight back onto it.

Hosted quick start

Use Node.js 22.22+ within Node 22, or Node 24, and run this in your app's repository:```bash
npx pincushion-mcp setup


The guided wizard signs you in to Pincushion, asks you to review the project URLs,
and explicitly choose an editor. It writes supported project MCP configuration when
safe; Manual, Skip, or a conflicting existing configuration can leave it unchanged.
Restart the editor and ask your agent to call `get_project_context` and
`get_actionable_pins`; configuration written alone does not verify a connection.
If you already have a critique report, run its `npx pincushion-mcp claim `
command from the owning app's repository instead, then verify a real pin read.

Pincushion hosts cloud sync. The normal setup does not require your own Supabase
project, database table, or backend API key. Sign-in supplies the existing account's
cloud credentials automatically; `npx pincushion-mcp login` is the recovery command
if sign-in is missing or expired. Do not paste credentials into issue reports.

Setup connects feedback access; it does not generate an AI critique. Critique execution
requires a supported coding-agent workflow and real page captures. A public report
lets reviewers add pins without an extension. Install the
optional Chrome extension only when you want
to place feedback directly on other web pages.

Use the maintained MCP setup guide for exact Claude Code,
Cursor, VS Code, and Codex commands, and the public documentation
for current plan limits and troubleshooting. The optional legacy self-host connection flags below are not prerequisites for
hosted setup.

## Claude Code plugin

This repository is also a Claude Code plugin marketplace. In Claude Code:```
/plugin marketplace add jcooley8/pincushion-plugin
/plugin install pincushion@pincushion

The plugin starts the Pincushion MCP server for the open project
(npx -y pincushion-mcp --project-dir --cloud-sync) and adds
/pincushion:pins, /pincushion:my-pins, /pincushion:implement,
/pincushion:resolve, /pincushion:feedback-summary and /pincushion:setup.
Run npx pincushion-mcp login once if the server reports local-only mode.

CLI Flags```bash

npx pincushion-mcp [flags]


| Flag | Description | Default |
|------|-------------|---------|
| `--project-dir PATH` | Root directory containing `.feedback/` | Current working directory |
| `--sync-url URL` | Optional legacy self-host adapter endpoint; not hosted setup | None |
| `--api-key KEY` | Optional legacy self-host adapter credential; not hosted setup | None |
| `--license-key KEY` | Explicit account credential override; normal hosted setup uses sign-in | Signed-in account when available |
| `--cloud-sync` | Turn hosted cloud sync on explicitly; it also turns on by itself when a signed-in account or `PINCUSHION_LICENSE_KEY` is present | On when an account is found |
| `--rest` | Enable REST API mode | Disabled (uses MCP/stdio) |
| `--port PORT` | Port for REST API server | 3456 |

### Examples

**Local project:**```bash
npx pincushion-mcp --project-dir /path/to/project

Optional legacy self-host adapter (not hosted Pincushion setup):```bash
npx pincushion-mcp \
--project-dir /path/to/project \
--sync-url https://abcd1234.supabase.co/api \
--api-key sb_project_key_abc123...


**REST API server:**```bash
npx pincushion-mcp --rest --port 8080

Tools

get_annotations

Retrieve annotations from .feedback/. Filter by page, component, or status.

Parameters:
- pageUrl (string, optional) — Filter by page URL (partial match)
- componentName (string, optional) — Filter by LWC component name
- status (string, optional) — Filter by open, in-progress, or resolved

Example:```javascript
await mcp.callTool('get_annotations', {
componentName: 'wmlHomePage',
status: 'open'
});


### `search_annotations`

Full-text search across all annotations, comments, selectors, and tags.

**Parameters:**
- `query` (string, required) — Search term

**Example:**```javascript
await mcp.callTool('search_annotations', {
  query: 'button label'
});

get_feedback_summary

High-level rollup of all feedback: counts by status, priority, page, and component.

Example:```javascript
await mcp.callTool('get_feedback_summary', {});


### `get_component_feedback`

Get all feedback for a specific LWC component with a plain-language summary.

**Parameters:**
- `componentName` (string, required) — LWC component name

**Example:**```javascript
await mcp.callTool('get_component_feedback', {
  componentName: 'wmlHomePage'
});

resolve_annotation

Mark an annotation as resolved after fixing the issue.

Parameters:
- annotationId (string, required) — Annotation ID
- comment (string, optional) — Resolution message
- resolvedBy (string, optional) — Name to attribute resolution (default: "AI Agent")

Example:```javascript
await mcp.callTool('resolve_annotation', {
annotationId: 'ann_abc123',
comment: 'Updated button label in line 42 of wmlHomePage.js'
});


### `add_agent_reply`

Add a reply to an annotation thread (e.g., ask clarifying questions).

**Parameters:**
- `annotationId` (string, required) — Annotation ID
- `body` (string, required) — Reply message
- `author` (string, optional) — Author name (default: "AI Agent")

**Example:**```javascript
await mcp.callTool('add_agent_reply', {
  annotationId: 'ann_abc123',
  body: 'Is this button in the main navigation or sidebar?'
});

fix_and_resolve

Combine fixing code and marking an annotation as resolved in one call. Optionally records commit / branch / PR metadata so the dashboard can backlink to what shipped.

When cloud sync is on, the response also says whether the cloud took the update: cloudSynced is true or false, and a false comes with cloudSyncError (license_rejected, server_error or network_error), the HTTP cloudSyncStatus when there is one, and a cloudSyncHint. A license_rejected means the pin was resolved locally but the extension still shows it READY: run npx pincushion-mcp login and resolve it again. In local-only mode the response is unchanged.

Parameters:
- annotationId (string, required) — Annotation ID
- fixDescription (string, required) — Description of the fix
- filePath (string, optional) — File where fix was applied
- lineNumber (number, optional) — Line number of the fix
- commitSha (string, optional) — Commit SHA that landed the change
- branchName (string, optional) — Branch the commit was made on
- prUrl (string, optional) — Pull request URL (GitHub/GitLab/Bitbucket; shape-validated)

Example:```javascript
await mcp.callTool('fix_and_resolve', {
annotationId: 'ann_abc123',
fixDescription: 'Updated button label to match design spec',
filePath: 'src/components/wmlHomePage.js',
lineNumber: 42,
commitSha: 'abc123def456',
branchName: 'pincushion/checkout-fix',
prUrl: 'https://github.com/acme/app/pull/142'
});


### `get_implementation_packet`

Fetch a single implementation packet for one page URL — selector list, full pin payloads, suggested branch name, and traceability config. Use when an agent wants to batch-fix one page in a single branch.```javascript
await mcp.callTool('get_implementation_packet', { pageUrl: '/checkout' });

assign_pin_to_agent

Dispatch a pin straight to your local coding agent. Promotes the pin to ready if not already, marks pending_implementation, and writes a .feedback/.agent-queue/.json trigger file that agent-loop.mjs picks up and shells out to Cursor / Claude Code / Codex.```javascript
await mcp.callTool('assign_pin_to_agent', { annotationId: 'ann_abc123' });


### `link_pin_deploy`

Attach a deploy URL to a resolved pin. Typically called by the deploy-hook edge function once production includes the fix, but available manually too.```javascript
await mcp.callTool('link_pin_deploy', {
  annotationId: 'ann_abc123',
  deployUrl: 'https://acme-app.vercel.app'
});

record_pin_verification

Write Pincushion AI's post-deploy verdict back to the pin. Called by the critic agent after /critique-latest-deploy runs against a fresh deploy.```javascript
await mcp.callTool('record_pin_verification', {
annotationId: 'ann_abc123',
status: 'verified', // or 'regressed' or 'inconclusive'
notes: 'Button matches the primary token. No regression on adjacent CTAs.'
});


### `get_time_to_fix_metrics`

**Pro/Team feature** — Free callers get sample size + upgrade hint. Median + p25/p75 of pin-to-resolve duration, with a 5-pin minimum so the metric is never noise.```javascript
await mcp.callTool('get_time_to_fix_metrics', { scope: 'project', projectId: 'pc_proj_abc' });
// → { sampleSize, thresholdMet, median, p25, p75, medianHuman, ... }

get_setup_instructions (NEW)

Get setup and configuration instructions for all supported agents.

Example:```javascript
await mcp.callTool('get_setup_instructions', {});


### Slack and Microsoft Teams integrations

Pincushion can notify Slack or Microsoft Teams through project-scoped incoming webhooks. The defaults are intentionally quiet and Figma-inspired: notify when a pin is ready for implementation, when someone is @mentioned, and when a collaborator adds follow-up on work already being handled. Every newly dropped pin and every resolution are opt-in events.

**Recommended use cases:**
- Developer channel: `pin_ready` and `follow_up`
- Design or PM channel: `mention` and optionally `resolved`
- Launch or QA channel: `pageUrlPatterns` plus `pin_ready`, `follow_up`, and `resolved`
- Temporary incident channel: enable a focused subscription, then pause it after the ship window

**Example:**```javascript
await mcp.callTool('configure_collaboration_integration', {
  projectId: 'my-project',
  provider: 'slack',
  webhookUrl: 'https://hooks.slack.com/services/...',
  targetLabel: '#product-feedback',
  events: ['pin_ready', 'mention', 'follow_up'],
  pageUrlPatterns: ['staging.example.com/checkout'],
  sendTest: true
});

For Slack, use create_slack_install_link when the hosted Slack app secrets are configured. It returns an Add-to-Slack URL; after approval, Slack returns the incoming webhook and Pincushion stores it automatically.

Use list_collaboration_integrations to audit configured destinations, remove_collaboration_integration to disconnect one, and preview_collaboration_notification to see the payload shape before adding a real webhook. Webhook URLs are stored server-side and returned only as masked values.


Auto-Agent Loop (Optional)

For agents that don't watch the file system (Claude Code, Cursor, generic),
agent-loop.mjs polls .feedback/.agent-queue/ and dispatches new pins
to the configured agent automatically.```bash

from inside the pincushion-mcp directory

npm run agent-loop -- --project-dir /path/to/your/project

or directly

node agent-loop.mjs --project-dir /path/to/your/project [--agent claude-code|cursor|generic] [--interval 3000]


The bridge (server.js) writes one trigger file per approved pin into
`.feedback/.agent-queue/`. The loop reads them, builds a prompt with
the pin's thread + element selector, and shells out to the chosen agent.
The agent uses MCP tools (`claim_pin` → fix → `fix_and_resolve`) and
the queue file is removed when the pin closes.

`detectAgent()` auto-detects `claude` or `cursor` on the `PATH`; falls
back to `generic` (writes the prompt to `.feedback/.agent-prompt` and
stdout). Run with `--interval 3000` to control poll cadence.

---

## Local File Structure

The server reads annotations from `.feedback/` in your project:```
.feedback/
├── annotations/
│   ├── example-com-login.json
│   ├── example-com-dashboard.json
│   └── ...
└── index.json

Each annotation file contains:```json
{
"pageUrl": "https://example.com/login",
"pageTitle": "Login",
"annotations": [
{
"id": "ann_abc123",
"status": "open",
"priority": "high",
"tags": ["design", "accessibility"],
"createdAt": "2026-03-19T10:30:00Z",
"element": {
"lwcComponent": "wmlLoginForm",
"selector": ".login-button",
"textContent": "Sign In"
},
"thread": [
{
"author": "Design Team",
"timestamp": "2026-03-19T10:30:00Z",
"body": "Button label should say 'Sign In' not 'Login'",
"type": "comment"
}
]
}
]
}


---

## Optional legacy self-host connection

The `--sync-url` and `--api-key` flags select a separate legacy sync adapter.
They are not needed to use Pincushion's hosted service. The older instructions to
create a Supabase project and annotations table are not a supported hosted
onboarding path; simply creating those resources does not establish a compatible
backend. Use this adapter only with an already compatible operator-managed endpoint.
See hosted setup for the default path and
current pricing for account plans.

---

## Troubleshooting

### "Module not found" error

Make sure you have a supported Node.js runtime: Node 22.22+ within Node 22, or Node 24:```bash
node --version

Install dependencies:```bash
npm install @modelcontextprotocol/sdk


### Annotations not appearing

Check that `.feedback/` exists in your project directory:```bash
ls -la .feedback/

If it doesn't exist, create it and add some test annotations, or the extension will create it when you pin your first feedback.

Hosted feedback not appearing

Run npx pincushion-mcp login if sign-in is missing or expired, then rerun setup
in the intended repository. Confirm the acknowledged project and registered page
URLs, restart the selected editor, and ask it to read get_project_context and
get_actionable_pins. Preserve existing project identity; do not create a replacement
project or backend to work around an access failure. See the
maintained troubleshooting guide.

Agent cannot find the server

Rerun npx pincushion-mcp setup from the intended repository, preserving the
acknowledged project identity and existing configuration. Follow the
maintained editor setup guide if setup leaves a conflicting
configuration unchanged. Restart the selected editor and verify a real project/pin
read. Avoid replacing a project-scoped configuration with an unscoped global entry.


Development

Clone the repository and install dependencies:```bash
git clone https://github.com/jcooley8/pincushion-plugin.git
cd pincushion-plugin
npm install


Run the server:```bash
npm start

Or with test data:bash npm start -- --project-dir ./test-feedback


License

MIT License. See LICENSE file for details.


Support

  • Issues: GitHub Issues
  • Docs: Pincushion Documentation

Changelog

v1.0.0 (March 2026)

  • Initial release
  • Support for Cursor, Claude Desktop, Claude Code, VS Code, Windsurf, Antigravity
  • Local .feedback/ file support
  • Supabase remote sync
  • REST API wrapper for non-MCP clients
  • New tools: fix_and_resolve, get_setup_instructions

常见问题

8 问
答案由本站依据条目字段整理,不引用上游原文
pincushion-plugin 怎么安装?

在客户端按下面的安装命令接入:npx pincushion-mcp setup。对应的包名是 pincushion-mcp。装完在客户端配置里引用同一个包或端点。

pincushion-plugin 怎么在 Claude Desktop、Claude、Cursor 里配置?

它的传输方式是stdio · 本地进程,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

pincushion-plugin 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

pincushion-plugin 需要 API Key 吗?

资料标注需要API Key认证,使用前准备好对应凭据,并按最小权限配置。

pincushion-plugin 支持哪些客户端?

资料标注支持 Claude Desktop、Claude、Cursor、VS Code、Windsurf、Codex;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

pincushion-plugin 能提供哪些能力?

资料显示能力标签包括 Local Deployment、MCP Server、AI Coding Agent、JavaScript、Visual Feedback、Code Fixing。具体工具清单以官方文档为准。

pincushion-plugin 需要什么运行环境?

运行时为 node;属于本地进程,需要在本机装好运行环境。

pincushion-plugin 是本地还是远程?

它是本地进程(stdio),由客户端在本机拉起。传输方式为stdio · 本地进程。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2026-10-08。