MCP Overview
Understand Kommon Poll MCP, what it exposes to AI clients, when to use it, and how it differs from the API and Kommon Poll application.
Connect Kommon Poll to AI clients so they can analyze authorized saved searches, overview metrics, and mentions through natural-language workflows.
Kommon Poll MCP is a live bridge between an AI assistant and the Kommon Poll data you are authorized to access.
MCP stands for Model Context Protocol. In practical terms, it lets an AI client call Kommon Poll tools during a conversation instead of relying on pasted exports or static files.
The AI client decides when to invoke Kommon Poll based on:
- The user's request.
- The available MCP tool descriptions.
- The permissions granted during authorization.
- The client mode and tool-use settings.
Kommon Poll MCP should be treated as an AI workflow integration. It is best when a user wants to ask questions, generate reports, inspect examples, compare signals, or create reusable analyst workflows.
What You Can Do
With Kommon Poll MCP, supported AI clients can:
- Discover saved searches and projects available to the authenticated user or team.
- Retrieve overview analytics for a saved search.
- Retrieve and inspect individual mentions.
- Apply supported filters for date, time, platform, author, hashtag, domain, tag, subtopic, intent, sentiment, subjectivity, influence, demographics, and related fields.
- Combine Kommon Poll data with the AI client's writing, reasoning, files, research, charts, presentations, email, calendar, code, or other connectors where supported.
- Turn natural-language questions into repeatable analysis workflows without manually exporting data.
The MCP integration should document each available tool as read-only or write-capable. If write tools are introduced later, they should be clearly separated from read-only tools and require explicit user confirmation.
MCP vs API vs Kommon Poll
| Use | Best option | Why |
|---|---|---|
| Ask conversational questions about brand, campaign, competitor, or customer signals | MCP | The AI client can choose Kommon Poll tools and explain findings in natural language. |
| Generate AI-assisted reports, briefs, or investigation summaries | MCP | The AI can combine overview analytics, mention evidence, and report writing. |
| Build deterministic backend services, dashboards, ETL, or custom software | API | Your application controls every request, response, retry, and storage pattern. |
| Explore charts, configure saved searches, adjust queries, or monitor daily activity | Kommon Poll application | The product interface is built for interactive setup and review. |
| Run scheduled server-side jobs with strict repeatability | API | A backend job is easier to control, test, log, and retry deterministically. |
Use MCP when the AI's reasoning and tool choice are part of the workflow. Use the API when the integration should behave the same way every time.
Prerequisites
Before connecting Kommon Poll MCP, confirm:
- You have a Kommon Poll account.
- You have access to the required team or workspace.
- MCP access is enabled for your package or plan, if feature-gated.
- At least one saved search or project exists.
- Your AI client supports MCP or Kommon Poll connectors.
- Your AI client plan and admin policy allow custom tools or connectors.
- You have permission to authorize the AI client for the relevant Kommon Poll team.
Basic Workflow
- Add Kommon Poll as an MCP server or connector in the AI client.
- Start the Kommon Poll authorization flow.
- Sign in to Kommon Poll.
- Select the team or workspace the AI client should access.
- Review and approve the requested scopes.
- Return to the AI client and verify that Kommon Poll tools are available.
- Ask the AI client to list accessible Kommon Poll projects.
- Choose a saved search and run an overview or mention workflow.
- Ask follow-up questions using the same saved search, date range, and filters.
Choose A Page
| Goal | Start here |
|---|---|
| Connect an AI client and understand OAuth | Connect And Authenticate |
| Understand available tools and data | MCP Tools Reference |
| Connect ChatGPT | ChatGPT Setup |
| Connect Claude | Claude Setup |
| Connect Gemini | Gemini Setup |
| Connect Microsoft Copilot | Microsoft Copilot Setup |
| Connect developer MCP clients | Other MCP Clients |
| Generate reports, briefs, and investigations | MCP Recipes |
| Create reusable prompts, skills, or agents | Prompts, Skills, And Agents |
| Schedule or automate workflows | Scheduling And Automation |
| Combine Kommon Poll with other tools | Cross-Connector Workflows |
| Secure MCP use and improve reliability | Security And Best Practices |
| Resolve common issues | Troubleshooting And FAQ |
Recommended Starting Stack
For most teams, start with:
- One authorized AI client connection per user and team.
- A short test prompt that lists accessible saved searches.
- A second test prompt that retrieves a saved-search overview.
- A third test prompt that retrieves a small page of mentions.
- Reusable instructions that force the AI to state the saved search, date range, filters, and limitations.
- A standard recipe for weekly reports or crisis checks.
- Clear revocation and reconnect steps for administrators.
Platform setup pages should be checked before publishing because AI client interfaces, plan names, transport support, and scheduled-task behavior change frequently.