Claudinho MCP Server

Community

Live 2026 World Cup scores in your terminal and AI agent

How to install ↓ Compare GitHub source repository ↗

Category: Other #10 of 230
Stars: ★ 31 Popular
Transport: stdio (local process)
Runtime: Node.js
Credentials: No credential needed
License: MIT
Last commit: 4 days ago
Tools: 9

70FMRS · B

Claudinho is a thoughtfully built unofficial World Cup scores MCP server: key-free, offline schedules, multilingual, with multiple surfaces (CLI/statusline/hook/MCP). Privacy-wise it collects no personal data, and security-wise it's stdio-only with two read-only outbound hosts. Great for fan-developers during the 2026 tournament; it relies on third-party live data, and its value fades after the event.

Strongest · Documentation 17/20 Weakest · Reliability 9/20

Reliability

9/20

Security and permissions

17/20

Maintenance

13/20

Documentation

17/20

Setup experience

14/20

Why each score

Reliability 9/20: Only the server manifest and README are available — no actual source code, test files, or CI workflow contents to inspect. The CI badge mentioned in the README is untrusted evidence and cannot be treated as reproducible proof. The manifest declares 9 read-only tools with an internally consistent description, so a happy path seems plausible, but there is no verifiable evidence of error handling, dependency control, or edge-case coverage. Per the static calibration rule (reliability capped at 12 without verifiable execution evidence), a low score is given.
Security and permissions 17/20: The server uses stdio transport only and requires no credentials; all environment variables (language, timezone, flavor, market-signal toggle) are optional display-only settings with no secrets or write access. The README/SECURITY.md describe a threat model of 'stdio only, no listener, no credentials, two read-only outbound hosts,' and the privacy policy states no personal data or telemetry is collected. No irreversible delete/pay/deploy/remote-command defaults are present, so no red line is triggered. Full marks are withheld because the actual source code cannot be inspected to confirm there is no undeclared network access or scope creep beyond the manifest.
Maintenance 13/20: The project is MIT-licensed, not archived, and shows a CI badge, npm provenance via OIDC trusted publishing, and a dedicated security-disclosure channel (GitHub Security Advisories). However, the version is 0.9.4 (pre-1.0), and the README explicitly self-describes as a 'solo, $0, organic fan project,' indicating a single-maintainer bus-factor risk and incomplete governance/versioning maturity, warranting a mid-range score.
Documentation 17/20: The README is thorough, covering installation across multiple clients (CLI, Cursor, Claude Code, other MCP clients), environment variable descriptions, an FAQ, the offline/online data boundary, troubleshooting for known terminal rendering quirks, and links to a privacy policy and security policy; the documented tool list matches the 9 tools in the manifest. Points are deducted for the absence of per-tool parameter schemas and for vague statements like 'rate limits respected' without concrete numbers or quotas.
Setup experience 14/20: The setup path is clear: npx one-liners, `claude mcp add` / `codex mcp add` commands, and standard stdio JSON config snippets for multiple clients (Cursor, Claude Desktop, Windsurf, Zed, VS Code), with few steps and multiple client examples. Per the static calibration rule, setup cannot exceed 15 without verifiable execution evidence (real CI workflows plus committed tests covering key paths), so a high but capped score is given within that ceiling.

Static review · not runListed 2026-09-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the network

Best for

Developers working in the terminal during matches who want scores at a glance
Claude Code or Cursor CLI users wanting a football statusline
MCP users who want key-free, zero-signup sports data

Not for

Users needing official or licensed tournament data (unofficial fan project)
Workflows requiring write operations or auth (read-only only)
Anyone seeking betting advice (market signals are informational only)

Required permissions

Read/write access to the local cache directory ~/.cache/claudinho
Read-only network requests to ESPN's public scoreboard during matches
Read-only requests to Polymarket public data when market signals are enabled
Modifying Claude Code / Cursor config files when installing the statusline and hook

Risks and side effects

Unofficial fan project; accuracy depends on third parties (ESPN, Polymarket) availability and quality
Live scores require network; offline you only get the bundled schedule and bracket structure
Prediction-market signals are informational only and must not be treated as betting advice
Solo-maintained $0 project; long-term maintenance after the tournament is uncertain

Setup

Before you start

Runtime：Node.js

Other optional settings (8)

CLAUDINHO_LANG optional Output locale: en, es, pt, or fr; optional, defaults to en.

CLAUDINHO_TZ optional IANA timezone (e.g. America/Mexico_City) for kickoff times; optional, defaults to system.

CLAUDINHO_FLAVOR optional Commentary flair level: off, subtle, or full; optional, defaults to full.

CLAUDINHO_MARKETS optional Set to off to hide prediction-market signals (informational only); optional.

CLAUDINHO_TEAM optional Scope to one team's matches (e.g. MEX) for CLI/statusline; optional.

CLAUDINHO_FLAGS optional Set off to use 3-letter codes instead of flag emoji, on to force flags; optional.

CLAUDINHO_CURSOR_META optional Set to auto to show a model + context line under the Cursor statusline score; optional.

CLAUDINHO_DEBUG optional Set to 1 to print data-provider failure diagnostics to stderr; optional.

Distributed via npm (@claudinho/mcp). Claude Code: install @claudinho/cli and run claudinho init claude (backs up ~/.claude/settings. first), or run claude mcp add claudinho -- npx -y @claudinho/mcp. Cursor: claudinho init cursor, or add {"mcpServers":{"claudinho":{"command":"npx","args":["-y","@claudinho/mcp"]}}} to ~/.cursor/mcp. or a project .cursor/mcp.. Codex CLI: codex mcp add claudinho -- npx -y @claudinho/mcp. Other MCP clients use the same stdio config. Restart your agent session. Optional env: CLAUDINHO_LANG (en/es/pt/fr), CLAUDINHO_TZ (IANA timezone), CLAUDINHO_FLAVOR (off/subtle/full), CLAUDINHO_MARKETS (off to hide market signals).

.mcp.json

{"mcpServers":{"claudinho":{"command":"npx","args":["-y","@claudinho/mcp"]}}}

Shown for Claude Code. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "claudinho": {
      "command": "npx",
      "args": [
        "-y",
        "@claudinho/mcp"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add claudinho -- npx -y @claudinho/mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Add {"mcpServers":{"claudinho":{"command":"npx","args":["-y","@claudinho/mcp"]}}} to your client config, restart the session, and confirm the 9 tools (get_today, get_live, get_match, get_next_fixture, get_standings, get_bracket, get_market_signal, get_share_snippet, get_team) appear in the tool list; asking about today's matches proves the connection.

Troubleshooting

Flags render as boxed letters (e.g. CH)? Your terminal (notably Warp) can't compose regional-indicator pairs; use CLAUDINHO_FLAGS=off for 3-letter codes
Statusline not updating? Restart the agent session; for local builds use granular init commands with --command
Cursor's beforeSubmitPrompt hook can't reliably inject context yet — the score-aware hook is Claude Code-only
Set CLAUDINHO_DEBUG=1 to print data-provider failure diagnostics to stderr

Things to try

Once connected, you can ask your AI assistant things like:

What matches are on today and what are the scores?
Which games are live right now?
Did Germany get through? Show me the knockout bracket.
When is Mexico's next fixture, kickoff time in Mexico City?

Tools 9

get_today read-only

Get today's matches and scores

get_live read-only

Get matches currently in play

get_match read-only

Query a single match by ID

get_next_fixture read-only

Get a team's next fixture (e.g. MEX)

get_standings read-only

Get group standings

get_bracket read-only

Get the knockout bracket

get_market_signal read-only

Read-only prediction-market favored line (Polymarket, informational only)

get_share_snippet read-only

Generate a shareable plain-text score/standings card

Show 1 more tools

get_team read-only

Resolve a team name to its code (e.g. "DR Congo" → COD)

Use cases

Asking your AI agent mid-task 'did Germany get through?' and getting the live result

Inline live scores in your Claude Code or Cursor CLI statusline

Checking group standings, the knockout bracket, and next fixtures with timezone support

Copying plain-text match/standings cards to your clipboard for group chats

Supported clients

Claude Code

Cursor CLIPartial support

Codex

Claude Desktop

Windsurf

Zed

VS Code

Listed from the project's documentation, not tested by this site.

Overview

Claudinho is an independent, open-source fan project (MIT, not affiliated with FIFA or Anthropic) delivering live 2026 men's World Cup scores, fixtures, and standings through an MCP server, a CLI, and Claude Code / Cursor CLI statuslines. No API key or signup required; all 104 fixtures ship bundled for offline schedules. Live data comes from ESPN's public scoreboard with attribution, and market signals from Polymarket (informational only). The MCP server runs over stdio with 9 read-only tools plus my_team / tournament_today prompts, supports en/es/pt/fr, and offers optional timezone and commentary-flavor settings.

Similar servers

Hevy MCP 77 · B

Manage your Hevy workout data from AI assistants via MCP.

★ 517 · Tools 26 Compare with this →

Firecrawl MCP Server 75 · B

Firecrawl's official MCP server for web search, scraping, and structured extraction for AI agents

★ 7.6k · Tools 11 Compare with this →

BioMCP 75 · B

One binary. One grammar. Evidence from the biomedical sources you already trust.

★ 659 · Tools 11 Compare with this →

Compartment 75 · B

Encrypted, fully offline long-term memory for AI agents

★ 583 · Tools 14 Compare with this →

Compare with other MCP servers

Source revision 91d797183606 Data synced 2026-10-19 Read the FMRS scoring method