OmniSignal

What OmniSignal MCP server does

The OmniSignal MCP server connects an AI assistant to a hosted workspace containing a brand profile, strategy outputs, market signals, source data, and a content calendar. Its tool set covers onboarding, profile and strategy retrieval, strategy generation, competitor-gap analysis, signal collection, calendar operations, data access, and credit information.

An agent can start by checking workspace readiness and gathering details such as goals, social handles, brand voice, and existing content. It can also analyze a website and connect supported platforms through the onboarding workflow. Once the profile is ready, the agent can request a strategy, inspect themes or SWOT results, and review the latest generated material.

Typical tasks include producing a market brief, comparing a competitor's coverage with the brand's own content, examining post ideas, finding current social posts for a query, and planning or revising upcoming posts. Calendar data can be retrieved by day, week, or post, searched, updated, and exported as CSV, ICS, or JSON.

How it works

The hosted service uses Streamable HTTP with one endpoint: https://mcp.omnisignal.ai/mcp. Requests carry an Authorization: Bearer header containing an API key created in the OmniSignal dashboard. The service is stateless, so the MCP client connects to the endpoint rather than starting a local process from this repository.

The OmniSignal MCP server exposes 38 tools grouped by function. Free tools handle many reads, onboarding actions, readiness checks, and credit queries. Generation and data operations consume credits and require the generate scope. Agents are instructed to check pricing and balance before paid calls, which helps expose the expected cost before an operation runs.

Some longer operations are asynchronous. Strategy generation and signal scanning return a job identifier that the client can poll with the related status tool. Failed asynchronous jobs are automatically refunded. Responses are split into smaller windows for model context; signal results use windows of 10, while calendar retrieval supports day-level responses and a week-oriented option.

Setup and configuration

Create an account at app.omnisignal.ai, open Profile, choose MCP / API access, and create an API key. The key is displayed once, can be revoked, and accounts can have up to 10 active keys. New accounts receive 50 starter credits.

Configure the remote endpoint in an MCP client using the URL and bearer header. Claude Desktop and Cursor use a Streamable HTTP entry with the endpoint and Authorization header. Claude Code can add the endpoint through its CLI, or use the repository's plugin workflow. For the plugin, set OMNISIGNAL_API_KEY; the plugin supplies the authentication header.

After connecting, ask the agent to set up the workspace. The onboarding flow checks readiness, collects profile information conversationally, analyzes the website, and begins the initial strategy process.

Tools and capabilities

The OmniSignal MCP server includes capabilities for:

• Checking and updating onboarding details, goals, social handles, brand voice, and content.
• Reading profiles, strategies, themes, SWOT analysis, and strategy status.
• Generating strategy, analyzing post ideas, and finding competitor content gaps.
• Scanning, listing, generating, dismissing, and snoozing market signals.
• Reading, searching, updating, varying, and exporting calendar posts.
• Collecting data, listing data points, viewing sources, and discovering sources.
• Checking credit balances and current tool pricing.

Paid calls use metered credits. The README states that one credit is valued at $0.15 on pay-as-you-go pricing, while subscription packages provide different effective rates. The hosted service is commercial even though the repository is licensed under MIT.