Text embeddings for AI agents — no API key, no account, no signup; pay per call in USDC on Base. POST JSON { input: string or string[] (max 64 per call) }. Model is jina-embeddings-v3, 1024 dimensions, retrieval.passage task. Returns { model, embeddings (one vector per input), dimensions, count, usage, latency_ms }. For agents building a vector index or doing semantic search without a Jina or OpenAI key.

Pull structured JSON out of messy text or a web page — no API key, no account, no signup; pay per call in USDC on Base. POST JSON { schema: { field: type, ... } (required), and text or url }. If url is given the page is fetched and stripped first; an LLM fills the schema and missing fields come back null. Returns { data, model, latency_ms }. For agents turning an article, listing, or product page into typed fields.

LLM chat completions for AI agents — no API key, no account, no signup; pay per call in USDC on Base. POST JSON { prompt } or { messages: [{ role, content }] }; optional model (OpenRouter slug), max_tokens (default 800, max 2000). Routes across a 5-model fallback chain so one unavailable model doesn't fail the call. Returns { model, content, usage, latency_ms }. For agents that need inference without holding an OpenAI or Anthropic key.

Mutable-risk report for a token — can the owner still turn it hostile? No API key, no account, no signup; pay per call in USDC on Base. GET ?address=0x... (required), chain=ethereum|base (default base). Direct on-chain reads only: eth_getCode, owner(), pendingOwner(), paused(), EIP-1967 slots. Returns mutable_risk_score, verdict, can_turn_hostile, time_to_rug, every owner power explained in plain English, and raw controls. For agents holding or about to buy a token.

Full wallet & token risk report — no API key, no account, no signup; pay per call in USDC on Base. GET ?address=0x... (required), chain=ethereum|base (default base), type=wallet|token (default wallet). Checks OFAC sanctions, scam/phishing lists, and honeypot/tax/mint signals. Returns risk_score 0-100, risk_level, a PROCEED|CAUTION|BLOCK verdict, flags[], reasons[] with severity, raw signals, and sources[]. For agents screening a counterparty before they transact.

Fetch any URL and get clean, LLM-ready page content — no API key, no account, no signup; pay per call in USDC on Base. Gets past common bot-blocking. Params: ?url=... (required), format=markdown|text|html (default markdown), max_chars=N (default 40000, max 120000). Returns { url, status, title, description, format, content, truncated, chars, latency_ms }; non-HTML URLs return raw content. For agents that need a page as text before summarising or extracting.

Live web search for AI agents — no API key, no account, no signup; pay per call in USDC on Base. GET ?q=YOUR+QUERY (required), count=N (1-20, default 10). Returns { query, results: [{ title, url, description, score }], provider, latency_ms }; score is 1.0 for the top result descending toward 0.1, so you can rank or threshold. Multiple search backends with automatic failover. For agents grounding an answer in current web data or gathering source URLs to scrape.