Revelica

com.revelica由 revelica.com 發布官方發布者v1.0.0更新於 Nov 2, 2026

Product intelligence for AI agents — query a Revelica workspace's ideas, specs, and research.

已驗證Streamable HTTP可網頁執行效率與工作流知識與記憶

概覽

AI 產生的概覽

遠端 Revelica MCP 伺服器，讓助理查詢、讀取、建立與更新 Revelica 工作區中的產品想法、規格與研究資料。

可以做什麼
  • 依條件搜尋或篩選工作區內容，回傳排序結果，指定類型時附上 schema 範本
  • 依 id 完整讀取單一產物或實體，或以點號欄位路徑瀏覽子樹
  • 建立新的產物與實體，內容依已註冊 schema 驗證
  • 以點路徑方式對欄位做局部更新
  • 列出並載入伺服器端的產品工作流程技能，例如朝目標規劃、設定賭注、撰寫規格或分析客戶訪談
適用情境
適合把客戶研究、競品分析、目標、賭注與產品規格放在 Revelica 的產品團隊，讓助理讀取這些脈絡並補上新的證據或計畫。常見情境包括把訪談逐字稿整理成快照、朝目標規劃工作，或把網址、檔案、筆記存成產品證據。
執行需求
需要一個 Revelica 工作區帳號。伺服器是遠端 Streamable HTTP 端點，採用 OAuth 2.0 與動態用戶端註冊，不需 API 金鑰，但用戶端須支援 OAuth。官方僅推出並測試 Claude 與 OpenAI 主機的外掛，其他支援 Streamable HTTP 的用戶端可直接連線。
安裝前請注意
伺服器可建立與更新工作區記錄，助理的寫入會改動真實產品資料。工具呼叫及其參數（包括你要求儲存的產品資料）會傳送到 Revelica 的伺服器。OAuth 與資料列層級安全機制將存取限制在本人工作區，資料處理受 Revelica 隱私政策與服務條款規範。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Revelica，將其新增到工作區。
  2. 連接 所需的 憑證。它們會被私密儲存，絕不會在公開頁面上展示。
  3. 為需要使用其工具的對話啟用該服務。

Web executable，透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式，或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南，先向你展示發布者、端點和登入方式，經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/com.revelica%2Frevelica/install.md，並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前，先告訴我它的發布者、端點和登入方式，等我確認。不要把 API 金鑰寫進命令或我會提交的檔案；問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。 客戶端會在首次連接時完成 OAuth 登入。

{
  "mcpServers": {
    "revelica": {
      "type": "http",
      "url": "https://api.revelica.com/mcp"
    }
  }
}

README

[Revelica]

Revelica Skills

Revelica's plugin package connects an agent to shared product context: customer research, competitive analysis, goals, bets and product specs. It declares the Revelica MCP server, a setup skill and three starting skills. The full workflows live on the server, which names the one for each job in its instructions and loads it with load_skill.

There is one plugin, plugins/revelica/, with a manifest for each host beside it. Each host's catalog sits at the repository root, where that host looks for it. Installation and OAuth support depend on the client.

Repository Structure

revelica/skills├── plugins/revelica/            # The plugin, shared by every host│   ├── .claude-plugin/          # Claude plugin manifest and icon│   ├── .codex-plugin/           # OpenAI (ChatGPT and Codex) plugin manifest│   ├── .mcp.json                # The Revelica MCP server│   ├── skills/                  # Starting skills (also slash commands in Claude Code)│   └── assets/                  # Brand icons├── .claude-plugin/marketplace.json   # Claude Code marketplace catalog├── .agents/plugins/marketplace.json  # Codex marketplace catalog├── server.json                  # MCP registry manifest (registry.modelcontextprotocol.io)├── scripts/package-openai.sh    # Exports the portable ZIP for OpenAI's plugin portal├── REVIEWING.md                 # Test setup for directory reviewers└── LICENSE                      # Apache-2.0

Installation by Platform

Claude Code

/plugin marketplace add revelica/skills/plugin install revelica@revelica

The plugin bundles the starting skills and the MCP server configuration. Authorize the server connection in your client before using workspace tools.

Cowork

Add https://api.revelica.com/mcp as a custom connector in your organization's settings. The server's instructions name the skill for each job, and list_skills lists the rest.

Codex

codex plugin marketplace add revelica/skillscodex plugin add revelica@revelica

In ChatGPT, install Revelica from the plugin directory once it is listed.

Other clients

Cursor, Gemini CLI and any other client that speaks Streamable HTTP with OAuth can connect the server directly (see Connecting the MCP Server Directly). We only ship and test plugins for Claude and OpenAI hosts. Release 1.10.1 removed the Cursor and Gemini manifests, which nobody was testing.

Starting prompts

The plugin adds a setup skill and three starting skills, the same starting points the Revelica app offers. In Claude Code each is also a slash command:

SkillWhat it does
/revelica:setup-revelicaSet up Revelica on your own product: your business, a goal and a researched customer, then a first result and a handover to planning. ChatGPT runs it from its Set up prompt after install; in Claude, ask to "set up Revelica".
/revelica:drive-product-outcomePlan the work toward our goal: read the goal, its measurements, the active bets and their evidence, then recommend the next action and plan it as work. Sets up anything the workspace is missing along the way.
/revelica:ingestionAdd insights to the knowledge graph: save a URL, file or note as product evidence, then choose the next useful work.
/revelica:interview-snapshotAnalyze a customer interview: turn a transcript into a snapshot and the customer problems worth solving.

What the plugin connects to and sends

The plugin runs no code on your machine. It declares one remote MCP server, https://api.revelica.com/mcp (Streamable HTTP), and bundles Markdown skills that start common work with it.

  • Sign-in: OAuth 2.0 with dynamic client registration, handled by your client. The plugin reads no credentials, tokens or files from your machine.
  • What it sends: the tool calls your agent makes (query, read, create, update, load_skill, list_skills) and their arguments, which carry the product data you ask it to save to your workspace.
  • What it receives: records from your own Revelica workspace, and skill instructions.

Nothing else is sent anywhere. See Privacy Policy for how Revelica handles your data.

Available MCP Tools

These tools are provided by the Revelica MCP server and callable from any skill:

ToolDescription
querySearch or filter the workspace by criteria. Returns ranked matches, plus the schema template when artifact_type is set.
readRead a single artifact or entity in full by id, or navigate to a subtree with a dotted field path.
createCreate new artifacts or entities. Content validated against registered schemas.
updateApply partial field-level updates via dot-path notation.
load_skillLoad a server skill's instructions, its readiness and a summary of the records it starts from. This does not itself execute a playbook.
list_skillsList the skills you can load, with each one's name, description and any required inputs.

All tools require OAuth authentication and enforce Supabase row-level security, so users only see their own workspace's data.

Skills

The product workflows are skills on the connected Revelica MCP server: drive-product-outcome, frame-bet, write-product-spec, design-experiment, product-for-coding-agents and others. The server's instructions say which one fits which job, and list_skills shows the full catalog for the connected user.

An Idea is the product spec. Its story map, structured document and Markdown are representations of the same spec.

Loading a skill lets the current agent follow the workflow using the available tools; it does not start a server-side playbook. These tool-based workflows do not require a host to implement the skills-over-MCP extension.

Updating

Product workflow instructions live on the Revelica server, so the app and connected agents discover the same canonical workflow. The package's four skills share the names and descriptions of the server skills they start, and load the full workflow from the server with load_skill, so the workflow and the workspace readiness it reports are always current. Release 1.9.0 removed the bundled use-revelica and product-for-coding-agents routers: every supported Claude host shows the server's instructions, which carry the same routing.

Release 1.10.0 moved the plugin into plugins/revelica/, turned the commands into skills so every host can use them, and added the OpenAI manifest. Bump the version in every manifest together, including server.json, and the SERVER_VERSION constant the MCP server reports (revelica_agents/app/mcp_server/server.py in scryfast). Name releases Revelica X.Y.Z and list the targeted platforms in the release notes. Versions belong to Revelica across hosts; platform-specific changes share that version. For example, Revelica 1.10.4 adds Codex plugin metadata and portable packaging for Codex and ChatGPT.

Package updates and server deployments are separate. A newly documented server workflow must be deployed before a connected client can load it. Installing this package does not deploy server changes.

Run bash scripts/package-openai.sh to build dist/revelica-openai-<version>.zip for OpenAI's submission portal. It exports the Codex manifest's listing fields under extensions.com.openai.interface in a portable root plugin.json, and exports .mcp.json as mcp.json with the Streamable HTTP transport name. Review cases and publication metadata are retained. The shared plugin's Claude and Codex compatibility manifests are not included in this public upload. Package validation does not complete the portal's reviewer access, demo, country targeting, scans, or attestations.

The exporter refuses to overwrite an existing archive. For a verification build, choose a new path: bash scripts/package-openai.sh --output /tmp/revelica-check.zip. ZIP entry timestamps and permissions are fixed, so builds with the same Python/zlib toolchain produce identical bytes. Run the packaging checks with python3 -m unittest discover -s scripts/tests -v.

Keep one private copy of the exact ZIP uploaded for review. dist/ stays ignored. Reviewer account details, recordings and submission notes belong outside this public repository. The public manifest contains reusable test cases; supply submission-specific workspace instructions and recording metadata privately when preparing an upload. A fresh export is not a replacement for the original upload.

The skills in plugins/revelica/skills/ carry the name, description and starting prompt of the server skill with the same name (scryfast revelica_agents/app/agents/revelica/skills/). Keep them in step when those change.

Connecting the MCP Server Directly

If you'd rather connect the server without the plugin, add it as a remote MCP server / custom connector:

Server URLhttps://api.revelica.com/mcp
TransportStreamable HTTP
AuthenticationOAuth 2.0 with dynamic client registration, so there's no API key to manage
PrerequisiteA Revelica workspace (sign up)

Server metadata is published at /.well-known/mcp/server-card.json.

Privacy Policy

Revelica's privacy policy covers what data is collected, how it is used and stored, third-party sharing, retention, and how to contact us. It is published at revelica.com/privacy. Terms of service are at revelica.com/terms.

The MCP server reads and writes only the product data in your own Revelica workspace. Every request is authenticated with OAuth and enforced by Supabase row-level security, so a connected agent can never see another workspace's data. The server does not query your chat history, conversation summaries, or files.

Support

License

Apache-2.0 © Revelica

About Revelica

Revelica is an AI-native product discovery platform. It helps product teams gather customer insights, perform competitive analysis, and experimentally validate ideas that feed into their AI development workflow.

🔗 Homepage with free Pro trial: https://revelica.com

🔗 Template library with playbooks, skills, and insight templates: https://revelica.com/templates

🔗 Free AI tools with no signup required: https://revelica.com/tools

來源：README.md，提交 372969e

工具

0
該伺服器需要登入，因此 SourceWeft 無法在沒有帳號的情況下讀取其工具，也未收錄。已連接帳號所能看到的內容可能因帳號及其權限而異。

版本歷史

1
  1. v1.0.0最新Sep 16, 2026

評分

已滿足適用檢查項的 85%

該評分衡量此收錄條目的完整程度，依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價，也不表示伺服器安全或目前可用：可連通性是帶日期的檢查結果，評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具15 分

    請讓伺服器在無憑證時回應 tools/list，或發布可在沙箱中無憑證啟動的軟體包。

  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分