Fantasy Premier League (FPL) WebMCP

Inspiration

Fantasy Premier League decisions involve constant trade-offs between form, fixtures, budget, and risk. We wanted to make that process collaborative, so a manager and an AI agent can inspect the same squad, make competing plans, and refine them together without touching the user’s real FPL team.

What it does

FPL Planner is a read-only transfer-planning workspace. Users load a public FPL team ID, create hypothetical transfer drafts, compare budgets and points hits, and change their starting XI, bench order, captain, and vice-captain.

The agent uses WebMCP tools to inspect squads and drafts, search players, create transfer scenarios, edit lineups, and query local Premier League history. Every transfer remains hypothetical.

How we built it

We built the app with Next.js, React, TypeScript, and the public FPL API.

The app fetches live player, squad, fixture, and entry data, then checks FPL rules locally, including budget, position quotas, club limits, transfer costs, and legal formations. We added a local dataset of four Premier League seasons so historical results can be queried without web search.

WebMCP exposes structured tools for draft reading, player queries, transfer scenarios, lineup changes, and historical match analysis.

Challenges we ran into

The main challenge was modelling FPL transfers accurately while keeping the interface simple. Every change can affect budget, free transfers, points hits, squad validity, captaincy, bench order, and draft history.

We also had to make human and agent changes work together. A draft should remain editable regardless of whether the user or agent created it.

Finally, we adapted the caching strategy for deployment. Local in-memory caching works during development, but a serverless deployment needs shared caching and explicit refresh behaviour.

Accomplishments that we're proud of

We created a genuinely collaborative planning experience rather than a chatbot that only gives recommendations.

The agent can work on the exact draft the user is viewing, while the user can inspect every decision, restore original players, make their own changes, and compare versions. The app also uses structured WebMCP tools instead of relying on UI scraping or ambiguous browser actions.

What we learned

We learned that the best agent experience is shared control, not full automation.

Clear, narrow tools with validation make agent actions more reliable and understandable. We also learned that live-data apps need thoughtful caching, refresh behaviour, and transparent constraints to build user trust.

What's next for Fantasy Premier League (FPL) WebMCP

We want to improve draft comparison, add richer transfer explanations, and give the agent more ways to evaluate fixture runs and long-term squad plans.

We would also expand historical analysis, support more advanced planning strategies, and make the app easier to use across more devices while preserving the same transparent, read-only workflow.

Built With

• codex
• vercel