OmniSignal

ai.omnisignal由 omnisignal.ai 發布官方發布者v1.0.0更新於 Nov 10, 2026

Live market intelligence & AI content strategy: trends, competitor moves, content calendar.

已驗證Streamable HTTP可網頁執行效率與工作流資料與分析商業與電商

概覽

AI 產生的概覽

OmniSignal 是遠端 MCP 伺服器，讓助理取得即時市場情報、競品動態，並為品牌產生 AI 社群內容行事曆。

可以做什麼
它提供 38 個工具，涵蓋導引設定、品牌資料與策略讀取、策略產生、市場訊號、內容行事曆管理，以及原始社群資料擷取。助理可以掃描所屬領域的趨勢與競品動作，產出附來源的簡報與差距分析，建立並編輯多週內容行事曆，產生 A/B 標題變體，並將行事曆匯出為 CSV、ICS 或 JSON。部分工具為非同步執行，會回傳任務 ID 供輪詢。
適用情境
適合需要為特定品牌持續進行社群內容規劃、競品研究或每週市場簡報的情境，且希望取得最新資料，而非模型訓練時的舊快照。對於一次性提問，或沒有社群帳號可追蹤的品牌，幫助不大。
執行需求
需要可連網的遠端 streamable HTTP 端點 OmniSignal 帳號，並在儀表板建立 API 金鑰，透過 Authorization 標頭以 Bearer omni_sk_... 形式送出。不需要本機執行環境或安裝套件；Claude Code 外掛會從 OMNISIGNAL_API_KEY 環境變數讀取金鑰。
安裝前請注意
Authorization 標頭攜帶具備 read 與 generate 權限的密鑰，金鑰僅顯示一次且可隨時撤銷。付費工具會消耗以 Stripe 計費的額度，呼叫會產生實際費用，月租方案的超額部分會自動以整數區塊計費。伺服器會蒐集品牌資料與社群帳號資訊，資料擷取會查詢 LinkedIn、Twitter 與 Instagram 內容。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 OmniSignal，將其新增到工作區。
  2. 連接 Authorization token 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/ai.omnisignal%2Fomnisignal/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。

{
  "mcpServers": {
    "omnisignal": {
      "type": "http",
      "url": "https://mcp.omnisignal.ai/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_TOKEN>"
      }
    }
  }
}

README

OmniSignal MCP Server

[omnisignal-mcp MCP server]

Live market intelligence and AI content strategy for your brand — straight into your AI assistant.

OmniSignal turns your brand profile into a data-driven social content operation: it collects fresh data from your niche (competitors, trends, audience), generates a multi-week content calendar, and surfaces daily signals (trends, competitor moves, opportunities). The entire backend is exposed as a remote MCP (Model Context Protocol) server — 38 tools your agent can call natively from Claude Code, Claude Desktop, Cursor, or any MCP-compatible client.

ChatGPT and Claude are trained on stale snapshots. Same model, same prompt — the only difference is whether your agent can see this week.

Server details

TransportStreamable HTTP (single endpoint, stateless)
Server URLhttps://mcp.omnisignal.ai/mcp
AuthAuthorization: Bearer omni_sk_... (API key from the dashboard)
Tools38 (onboarding, strategy, signals, calendar, data, credits)
DocsLive API docs · Markdown for agents
Websiteomnisignal.ai

Quickstart

1. Get an API key

Sign up at app.omnisignal.ai → Profile → MCP / API access → create a key (omni_sk_..., shown once). Keys carry read + generate scopes, up to 10 active keys, revocable anytime. New accounts get 50 free starter credits.

2. Connect your client

Claude Code (CLI):

bash
claude mcp add --transport http omnisignal \  https://mcp.omnisignal.ai/mcp \  --header "Authorization: Bearer omni_sk_YOUR_KEY_HERE"

Claude Code (plugin, from this repo):

/plugin marketplace add omnisignal-ai/omnisignal-mcp/plugin install omnisignal@omnisignal

Then set the OMNISIGNAL_API_KEY environment variable to your key — the plugin wires the header for you.

Claude Desktop (claude_desktop_config.json):

json
{  "mcpServers": {    "omnisignal": {      "type": "streamable-http",      "url": "https://mcp.omnisignal.ai/mcp",      "headers": { "Authorization": "Bearer omni_sk_YOUR_KEY_HERE" }    }  }}

Cursor (~/.cursor/mcp.json):

json
{  "mcpServers": {    "omnisignal": {      "type": "streamable-http",      "url": "https://mcp.omnisignal.ai/mcp",      "headers": { "Authorization": "Bearer omni_sk_YOUR_KEY_HERE" }    }  }}

Any MCP client: POST to the server URL with header Authorization: Bearer omni_sk_YOUR_KEY_HERE and Accept: application/json, text/event-stream.

3. Let the agent onboard you

Say: "Set up my OmniSignal workspace." The server ships with a built-in onboarding playbook — the agent checks readiness, collects your brand profile conversationally, analyzes your website, and kicks off your first strategy generation.

What you can do with it

  • Monday market brief — "What's moving in my market this week?" → fresh scan + a sourced brief of trends, competitor moves, and opportunities.
  • Battlecards / competitive response — sourced gap analysis of what a competitor covers that you don't, with immediate actions.
  • Weekly planning — walk this week's content calendar, playbooks included; A/B hook variants; mark posts published.
  • Export anywhere — the calendar as CSV / ICS / JSON into Notion, Google Calendar, or a scheduler.
  • Deep data pulls — fresh LinkedIn / Twitter / Instagram posts on any query when the agent needs raw material.

Tool overview (38 tools)

Free tools are always free; paid tools require the generate scope and credits. Agents are instructed (via the server's built-in instructions) to check credits_get_pricing / credits_get_balance before running paid calls.

GroupToolsPrice
Onboardingonboarding_get_readiness, onboarding_update_profile, onboarding_update_goals, onboarding_update_social_handles, onboarding_update_brand_voice, onboarding_update_content, onboarding_connect_platform, onboarding_analyze_websiteFree
Profile & strategy readsprofile_get, strategy_get_latest, strategy_list, strategy_get_status, strategy_get_swot, strategy_get_themesFree
Strategy generationstrategy_generate (async), strategy_analyze_post_idea, strategy_competitor_gap150 / 8 / 12 cr
Signalssignals_scan (async) + signals_scan_status (free), signals_list, signals_generate, signals_dismiss, signals_snooze, sources_list (free), sources_discover25 cr scan, 5 cr/list window, 10 cr generate/discover
Calendarcalendar_get_day, calendar_get_week, calendar_get_upcoming (free), calendar_get_post, calendar_search_posts, calendar_update_post (free), calendar_create_variants, calendar_export5 cr day / 15 cr week / 3 cr post / 3 cr search / 8 cr variants / 5 cr export
Datadata_summary (free), data_collect, data_points_list2–10 cr
Credits metacredits_get_balance, credits_get_pricingFree

Design notes: async jobs (strategy_generate, signals_scan) return a job id to poll and auto-refund on failure; responses are chunked for LLM context (signal windows of 10, calendar day-by-day with a week bulk discount); paid mutations are idempotent; per-user daily spend cap.

Pricing

1 credit = $0.15, metered to Stripe on every paid call.

PackagePrice/moCreditsEffective $/cr
Starter$25250$0.10
Creator$49550$0.089
Pro$991,200$0.083
Team$1992,600$0.077
PAYG—$0.15/cr, billed daily in arrears$0.15
  • 50 free starter credits on the free/PAYG tier.
  • Monthly packages never hard-stop: overage auto-bills in round-up blocks at the package rate.
  • Any MCP package also unlocks the full OmniSignal dashboard on the same account.

Links

  • Product site: omnisignal.ai
  • Dashboard: app.omnisignal.ai
  • Live MCP docs: /mcp/docs
  • This repo also contains server.json (Official MCP Registry manifest) and a Claude Code plugin marketplace manifest (.claude-plugin/marketplace.json).

License

MIT — see LICENSE. The hosted OmniSignal service itself is commercial; this repo contains listing metadata and documentation only.

來源：README.md，提交 729f1c6

工具

0
該伺服器需要 API 金鑰等憑證，因此 SourceWeft 無法在沒有憑證的情況下讀取其工具，也未收錄。

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