CompetLab for AI Agents
What this is
This is the agent-facing side of CompetLab — how you use your competitive intelligence from inside an AI assistant, rather than by writing code against the API. Where the developer platform is for programs you build (the REST API and the SDK), this is for AI agents you already use: connect one to your data and it can read your dimensions, competitors, alerts, and the Strategic Briefing in the flow of a conversation — or run a prebuilt skill that produces a finished analyst deliverable in one request.
Two surfaces
Works with any MCP client:
| MCP Server | Agent Skills | |
|---|---|---|
| What it is | A hosted Model Context Protocol server that exposes your data as tools | A suite of 13 prebuilt Claude Code skills |
| What you get | 33 callable tools any MCP client can discover and use | Finished deliverables — briefings, dossiers, battlecards, a CMO report |
| Use it to | Give an agent live, ad-hoc access to your data | Run a repeatable, expert-authored workflow with one request |
| Relationship | The foundation | Built on top of the MCP server |
The two work together: the MCP server is how an agent reaches your data; Agent Skills are prebuilt playbooks for turning that data into something useful.
Agent Skills run on top of the MCP server. Configure your MCP connection first, then add skills — they call its tools under the hood.
Which one do I want?
- Connecting an assistant to CompetLab — Claude, Cursor, VS Code, Windsurf, Cline — so it can read your data on demand? Start with the MCP Server.
- Want finished CI deliverables — a briefing, a competitor dossier, a sales battlecard, a full CMO report — from a single request? Install the Agent Skills (they need the MCP server configured first).
Building in code instead?
If you’re integrating CompetLab into your own backend or a script rather than an AI assistant, the developer platform — the REST API and the TypeScript SDK — is the path. Same data underneath; different caller.
Next steps
- MCP Server → — connect your agent; 33 tools; ready-to-paste client configs for every supported client.
- Agent Skills → — the 13-skill suite, how to install it, and a reference for every skill.
New to CompetLab? Create a free account and generate an API key — you’ll need one to connect the MCP server.
FAQ
What does "CompetLab for AI Agents" cover?
It covers the two ways to use CompetLab from inside an AI assistant rather than by writing code. The MCP Server is a hosted Model Context Protocol server that exposes your competitive intelligence as 33 callable tools, so any MCP-capable client — Claude, Cursor, VS Code, Windsurf, Cline — can read your data in a conversation. Agent Skills are 13 prebuilt Claude Code workflows that turn that data into finished deliverables like briefings, dossiers, and battlecards. The skills run on top of the MCP server.
What's the difference between the MCP Server and Agent Skills?
The MCP Server is the connection — it gives an agent live, ad-hoc access to your CompetLab data as tools it can call. Agent Skills are prebuilt playbooks on top of that connection: instead of asking an agent to figure out a good competitor dossier from raw tools, you install a skill that already knows how to produce one. The MCP server is the foundation; the skills are workflows built on it, and they call its tools under the hood — so you configure the MCP connection first, then add skills.
How is this different from the developer platform?
Same data, different caller. The developer platform (the REST API and the TypeScript SDK) is for programs you write — a backend service, a scheduled job, an internal tool. The AI Agents surfaces are for AI assistants you use — you connect one via the MCP server, or run a prebuilt Agent Skill. Under the hood the MCP server sits on top of the same REST API, so anything you can reach one way you can reach the other; you pick the surface that matches who's calling.