Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add edycutjong/cuporacle-mcp --skill cuporaclegit clone --depth 1 https://github.com/edycutjong/cuporacle-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/edycutjong/cuporacle-mcp/cuporacle)<a href="https://agentmods.dev/skills/edycutjong/cuporacle-mcp/cuporacle"><img src="https://agentmods.dev/badge/skills/edycutjong/cuporacle-mcp/cuporacle/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/edycutjong/cuporacle-mcp/cuporacle"><img src="https://agentmods.dev/badge/skills/edycutjong/cuporacle-mcp/cuporacle.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00098 | $0.01162 |
| Opus 5 | $0.00049 | $0.00581 |
| Sonnet 5 | $0.00020 | $0.00232 |
| Haiku 4.5 | $0.00010 | $0.00116 |
Grade A, and why
cuporacle scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CupOracle — World Cup intelligence + autonomous paid edge
CupOracle is an MCP server (cuporacle-mcp). It gives you 8 tools, 2 resources
(wc://bracket, wc://ledger) and this Skill. Four tools are free data; one
(wc_edge) spends real USDC via x402 and must be used responsibly.
Tool selection — question → tool
| The user asks… | Call | Notes |
|---|---|---|
| "What matches are on / tonight?" | wc_fixtures(date?) |
Omit date for the next upcoming window. |
| "What's the score / is it live?" | wc_live(matchId) |
Get the id from wc_fixtures first. |
| "What are the odds / who's favorite?" | wc_odds(matchId) |
Consensus decimal odds + de-vigged probabilities. |
| "Who's through / show the bracket?" | wc_bracket() |
Or read the wc://bracket resource. |
| "Is it worth a bet? Find me an edge." | wc_edge(matchId) |
Pays ~0.05 USDC. See spend policy below. |
| "Prove that payment / check this receipt." | receipt_verify(txHash) |
Reads the tx on Injective EVM. |
| "How do I fund the agent wallet?" | wallet_fund_guide(chain?) |
CCTP runbook. Then use the Injective MCP server. |
| "What have you spent?" | wc_spend_ledger() |
Session total vs cap + every receipt. Or wc://ledger. |
Always resolve a matchId with wc_fixtures before calling id-based tools.
Spend policy (READ BEFORE CALLING wc_edge)
wc_edgepays for itself via x402 (no API key, no account). Each call costs roughly 0.05 USDC. There is a per-session cap (CUPORACLE_MAX_SPEND, default 0.50 USDC).- Never raise your own cap. If a call returns
SPEND_CAP_HIT, STOP and ask the human to approve a higher cap (they restart the server with a new value). Do not retry in a loop. - Pass
maxSpendto bound a single call (e.g.wc_edge(matchId, maxSpend: 0.05)). - Always cite the receipt. When you present a paid edge, quote
receipt_txand offerreceipt_verify(txHash)so the user can confirm it on Injective. Format: "Edge: … · Paid 0.05 USDC · Receipt0x…(verify with receipt_verify)." - If LineLock is unreachable or the wallet is unfunded,
wc_edgedegrades to freewc_oddsand says "no vetted edge available". Never invent an edge or a receipt. Present the free odds honestly instead. - To preview the payment logic without spending, call
wc_edge(matchId, dry_run: true)— it parses the 402 quote and checks the cap but makes no payment.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 76 lines · 98 tokens per session scan A a51cb95184d3
cuporacle is a skill published in the GitHub repository edycutjong/cuporacle-mcp (0 stars, last pushed 5d ago), licensed MIT. It adds 98 tokens to every session and 1,162 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
worldcup-linelock
Buy and audit World Cup betting edges from LineLock over Injective x402. Use when the user wants the next knockout pick (model probability vs market odds + a conviction staking ladder), wants to pay per-call in USDC without an API key, or wants to audit a tipster's settled record by closing-line value (CLV). Handles…
proofline-match-verifier
Verify football match events and result readiness with Proofline evidence, signed x402 proof packets, and the latest Injective testnet settlement commitment. Use when an agent must check a score, event, final result, paid proof, or settlement gate without confusing scheduled, delayed, replay, and live data or…
workspace-dispatch
Single-agent mission orchestrator. Decomposes a mission into tasks, spawns one worker per task using the default model, verifies exit criteria, and chains tasks with retry. No critic pattern — each worker self-verifies. Simple, fast, works with any model config.
integration
Discover, install, and configure Vercel Marketplace integrations (databases, auth, logging, etc.). Use when adding third-party services to a Vercel project.
v0
Generate and iterate on UIs/apps with Vercel v0 via the v0 Platform API. Use when building a frontend with v0, calling the v0 API/SDK, or spending v0 credits.
vercel
Deploy, manage, and develop projects on Vercel from the command line.