pronto.stream is hosted on Hetzner.
| Technology | Version | First seen |
|---|---|---|
| Frontend Build Tools & CSS2 | ||
| – | 8 Nov 2026 | |
| – | 8 Nov 2026 | |
| Web Servers, Languages & Backend Frameworks1 | ||
| – | 8 Nov 2026 | |
| CDN & Hosting1 | ||
| – | 8 Nov 2026 | |
| Fonts & Icon Fonts1 | ||
| – | 8 Nov 2026 | |
Registration from the registry's RDAP record, the certificate the site serves, and its rank in public traffic lists.
The crew hasn’t read this site’s colors yet. We read a palette when we capture a site’s screenshot, busiest sites first. Meanwhile, browse palettes from other sites.
eforward2.registrar-servers.comeforward3.registrar-servers.comeforward1.registrar-servers.comeforward4.registrar-servers.comeforward5.registrar-servers.comdns2.registrar-servers.comdns1.registrar-servers.com_dmarc.pronto.stream. TXT v=DMARC1; p=none;v=spf1 include:spf.efwd.registrar-servers.com ~allgoogle-site-verification=VqWaiYhFKD7kxUJbp7uT-e0iBEW2TC-1_GLwPB5sBaw# pronto.stream # # The sitemap below is an index. Its children are sitemap-core.xml and # sitemap-hubs/N.xml. User-agent: * Allow: / Allow: /seo/ Allow: /mcp Allow: /llms.txt Allow: /llms-full.txt Allow: /openapi.json Allow: /server.json Allow: /terms Allow: /.well-known/ # The authenticated API. Nothing here is indexable and every path answers 401 to # an anonymous fetch, so crawling it spends budget to collect errors. Disallow: /v1/ Disallow: /api/ # ============================================================================== # 1. Frontier AI Training & Evaluation Crawlers (US & Global) # ============================================================================== User-agent: GPTBot Allow: / User-agent: ClaudeBot Allow: / User-agent: anthropic-ai Allow: / User-agent: GrokBot Allow: / User-agent: Google-Extended Allow: / User-agent: Applebot-Extended Allow: / User-agent: Amazonbot Allow: / User-agent: meta-externalagent Allow: / User-agent: CCBot Allow: / User-agent: Diffbot Allow: / # ============================================================================== # 2. Real-Time AI Search, Citation & User-Browsing Agents (US & Global) # # xAI publishes several tokens and reporting on which one actually arrives is # inconsistent, so all of the documented spellings are named rather than one. # A robots record costs a line; a crawler that finds no record for itself and # falls through to `*` is indistinguishable, in our own logs, from one that # never came. What settles it is the User-Agent recorded at the edge — see # `crawlerAgents` in internal/service/analytics/service.go, which counts them. # ============================================================================== User-agent: OAI-SearchBot Allow: / User-agent: ChatGPT-User Allow: / User-agent: Claude-SearchBot Allow: / User-agent: Claude-User Allow: / User-agent: xAI-Grok Allow: / User-agent: Grok-DeepSearch Allow: / User-agent: xAI-Bot Allow: / User-agent: Grok Allow: / User-agent: PerplexityBot Allow: / User-agent: GoogleOther Allow: / User-agent: YouBot Allow: / # ============================================================================== # 3. Chinese Frontier AI Crawlers & Real-Time Search Engines # ============================================================================== User-agent: DeepSeekBot Allow: / User-agent: deepseek-crawler Allow: / User-agent: Bytespider Allow: / User-agent: Baiduspider Allow: / User-agent: Baiduspider-render Allow: / User-agent: QwenBot Allow: / User-agent: KimiBot Allow: / User-agent: MoonshotBot Allow: / User-agent: ChatGLM Allow: / User-agent: GLM-Crawler Allow: / User-agent: MiniMaxBot Allow: / User-agent: Sogou spider Allow: / User-agent: 360Spider Allow: / User-agent: YisouSpider Allow: / # ============================================================================== # 4. European & Independent Alternative AI / Search Engines # ============================================================================== User-agent: MistralAI Allow: / User-agent: MistralBot Allow: / User-agent: AlephAlphaBot Allow: / User-agent: MojeekBot Allow: / User-agent: Qwantify Allow: / User-agent: DuckDuckBot Allow: / User-agent: BraveBot Allow: / User-agent: YandexBot Allow: / User-agent: SeekportBot Allow: / Sitemap: https://pronto.stream/sitemap.xml
# pronto.stream — Machine Readable Context for AI Agents & Search Engines
> pronto.stream is a real-time signal engine and Model Context Protocol (MCP) server. It fuses 96,814 RSS feeds across 230 countries and 99 structured API connectors into one cited wire, with 15 tools and 35 cross-domain fusion products, in a compact format that costs about 50% fewer prompt tokens than the same rows as indented JSON.
> Signals processed to date: 609,234,154 across 270 subject countries (lifetime counters, durable across restarts; see /seo/stats.json).
## What This Wire Is For
- **News and world state**: Every hazard, filing, outbreak and launch on one clock, each row carrying when it was observed and where it came from. global_news_attention ranks what the world is actually reading right now; news_response_latency measures how long a physical event takes to reach the press, joined on the gazetteer place the two happen to share rather than on the country they were filed under. Co-occurrence, not causation, and the row says so.
- **Macro and finance**: The published models, implemented faithfully and cited. recession_risk_12m forecasts; sahm_recession_indicator says whether a downturn has probably already started; they disagree on purpose and both are returned. taylor_rule_policy_gap prices the distance between the policy rate and the rule, macro_stress_index and sovereign_debt_dynamics carry the rest of the frame.
- **Quant and market microstructure**: insider_filing_cluster reads SEC Form 4 filings and reports when several insiders at one issuer transact inside the same window, which is a signal no single filing carries. fx_fixing_drift and fx_reference_rate_consistency compare independent central-bank reference rates against each other, so a rate that only one publisher believes is visible as a disagreement rather than as a number.
- **Science and research**: seismic_solution_consensus is the sharpest case for why a wire beats an API: several national agencies locate one earthquake and this reports how far apart their solutions are, in magnitude, depth and true three-dimensional hypocentre. Agencies routinely differ by half a unit while an event is still being located, which is a factor of five in energy. Alongside it, research_momentum tracks publication and citation velocity, space_weather_ops implements the NOAA scales, and neo_kinetic_energy states near-earth object energy in the units the literature uses.
- **Crypto**: crypto_cross_rate_dislocation reads the order books of several venues in several fiat currencies at once and reports where the implied cross rate disagrees with the direct one. It exists only because the venues are time-aligned on one wire; no single exchange API can compute it, because no exchange sees another's book.
- **Arbitrage and dislocation**: Three products measure a spread rather than a level: crypto_cross_rate_dislocation across venues, fx_reference_rate_consistency across central banks, and grid_carbon_arbitrage across the carbon intensity of neighbouring grids in the same hour. Each states its formula and cites the signals it came from, so the spread can be audited before it is traded on. None of them is advice, and a product with no live inputs says so instead of guessing.
## Target Market Taxonomy & Segment Value
- **AI Agent Market (LLM Tooling & MCP Clients)**: 15 native MCP tools over streamable HTTP/WebSocket; WebMCP browser runtime discovery; CWF token economics (saving 33-37% prompt tokens vs compact JSON, ~50% vs indented JSON); deterministic `src:` citations, source URLs, and ISO-8601 timestamps for inspectable source grounding; sub-microsecond Hermes hotplane reads with bounded context budgets.
- **Web3 Market (Crypto Protocols, DeFi & DAOs)**: Cross-domain crypto alpha (`crypto_cross_rate_dislocation`); multi-source correlation fusing mempool dynamics with SEC EDGAR filings and GDELT sentiment; off-chain event consensus before on-chain reaction; auditable `src:` provenance correlation IDs for automated DAO governance and compliance reporting.
- **Base Ecosystem (On-Chain Autonomous Agents & Builders)**: Native sensory signal wire for Base AI agents, Farcaster frames, Clanker tokens, and on-chain treasuries; agent-initiated USDC purchases through MCP on Base Mainnet (CAIP-2 `eip155:8453`, contract `0x833589fcd6edb6e08f4c7c32d4f71b54bda02913`); self-funding agent lifecycle without human credit cards. Technical guide: https://pronto.stream/seo/base-ai-agents-and-onchain-signal-infrastructure.html
## Quickstart & Public Verification (Zero Credentials Required)
The MCP endpoint is 100% publicly accessible without an API key, signup, or credit card.
1. **Instant Keyless Test (Terminal)**:
```bash
curl -s -X POST https://pronto.stream/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"get_latest_signals","arguments":{}}}'
```
2. **Agent / IDE Config (Claude Desktop, Cursor, Windsurf, LangChain)**:
```json
{
"mcpServers": {
"pronto": {
"type": "http",
"url": "https://pronto.stream/mcp"
}
}
}
```
3. **Lift Rate Limits & Access Commercial Alpha**:
- Human self-serve: https://pronto.stream/pricing
- Machine self-serve: Call `register_agent` to mint an identity, then use the MCP payment tools below.
## Pricing & Plan Limits
The smallest purchase is $1 of call credits: 10,000 calls at $0.0001 each, with 1200 calls a minute and 100 rows a response while the balance lasts, no expiry, and no charge for a call that returns an error. Builder ($1 a day, $3 a week, $5 a month) and Team ($5 a day, $20 a week, $50 a month) are periods; nothing renews automatically, and buying the same plan again before it ends extends it. Enterprise is a design partner engagement with no list price. An unauthenticated caller gets a free public preview with no key required.
- **Anonymous** (free): 30 tool calls/minute, 25 rows per response, the public alpha indicators.
- **Call credits** ($0.0001 a call, prepaid in top-ups of $1 to $100, no expiry, charged only for calls that return an answer, used whenever no plan is active): 1200 tool calls/minute, 100 rows per response, the public alpha indicators.
- **Builder** ($5/mo, or a $1 day pass or $3 week pass): 1200 tool calls/minute, 100 rows per response, the public alpha indicators.
- **Team** ($50/mo, or a $5 day pass or $20 week pass): 12000 tool calls/minute, unmetered rows per response, every alpha indicator including the commercial set.
- **Enterprise** (design partner — contact us): unmetered tool calls/minute, unmetered rows per response, every alpha indicator including the commercial set.
Payment rails: Self-serve checkout at https://pronto.stream/pricing or MCP open_payment and settle_payment with USDC on Base Mainnet. The x402-shaped document is for discovery; automatic x402 payment headers are not accepted. A response shortened by the row budget carries `NOTE|budget:N|requested:M|reason:context_budget`.
## Independent Proof & Verification
- **Audit Feed & Connector Counts (96,814 feeds, 99 connectors)**:
- Query MCP: call `list_data_sources` with `{"limit": 50}` to inspect live upstream endpoints and publisher licenses.
- REST & Sitemap: `GET https://pronto.stream/v1/pronto/sources`, `https://pronto.stream/seo/stats.json`, or `/sitemap.xml`.
- **Audit Token Savings (~50% vs indented JSON, ~33% vs compact JSON)**:
- Test datasets & benchmark suites: published at https://pronto.stream/seo/evidence-blocks.json.
- Reproduce independently using OpenAI's `cl100k_base` tokenizer (`tiktoken`):
`import tiktoken; enc = tiktoken.get_encoding('cl100k_base'); len(enc.encode(cwf)) / len(enc.encode(json_pretty)) -> ~0.50`
- CWF format specification: https://pronto.stream/seo/cwf-token-saver-format.html
## Working Without a Human
Four calls, none of which need a person. This is the whole path.
1. `register_agent` — mints an API key and an Ed25519 identity with no human present. Send the key as an Authorization header afterwards.
2. `get_plans` — the ladder as data: call credits, day and week passes, and plans, with a SETTLE line naming the asset, network and settlement address.
3. `open_payment` — quotes the exact amount, binds the purchase to the wallet you will pay from, and returns a SIGN line with the exact message to sign. For credits, pass `plan: "credits"` and `amount_usd` from 1 to 100; for a pass, `cycle: "day"` or `"week"`.
4. Transfer the USDC on Base Mainnet (chain 8453, CAIP-2 eip155:8453, token 0x833589fcd6edb6e08f4c7c32d4f71b54bda02913), then `settle_payment` with the transaction hash and a `signature`: the paying wallet's personal_sign (EIP-191) over the SIGN message with your lower-case hash filled in. Smart-contract wallets are checked with ERC-1271.
Why the signature: a transaction hash and the wallet it came from are both public once mined, so either alone would let anyone watching the chain redeem your transfer as theirs. Settlement is verified by reading the chain: the transfer must reach the settlement address, from the wallet you declared, for at least the amount owed, and be five blocks deep. One transaction pays for one purchase, once. A payment that is not yet deep enough settles on its own in the background; calling `settle_payment` again with the same hash reads the outcome. Your existing API key then resolves to what you bought with nothing to change on your side — you become your own principal, and no person holds the subscription.
USDC is the only accepted asset; there is no card rail. A person buys at https://pronto.stream/checkout: opening a checkout quotes an exact amount ending in a few millionths of a dollar unique to that checkout, which is what proves the payment is theirs, so it works from an exchange withdrawal too. Send exactly that amount within 72 hours, then paste the transaction hash. Machine-readable terms, in the x402 v2 form: https://pronto.stream/.well-known/x402
## Why the Format Is Compact
- The Cognitive Wire Format is a line-oriented text schema. It costs about 50% fewer prompt tokens than the same rows as indented JSON, and about a third fewer than compact JSON. Both figures are measured rather than estimated; the per-batch numbers are published at https://pronto.stream/seo/evidence-blocks.json. It exists because an agent polling a wire pays for every response it reads, twice: once to fetch and once to reason over.
- Every row carries its own citation handle. Cite the handle, pass it back to `get_signal` to re-read that row, or to `get_related_signals` for the rest of the story.
- Nothing is dropped silently: a truncated response says so, a withheld indicator names why, and a degraded source is listed so an absence reads as an outage rather than as calm.
## Release Stage
- Generally available (GA). pronto.stream is in full production General Availability as of September 2026, with over 609,234,154 lifetime signals processed across 270 subject countries.
- Disambiguation of 'Alpha': Historical reviews or crawler snapshots from August 2026 citing an 'Alpha stage' reflect early pre-GA preview testing. Furthermore, 'Alpha' in Pronto technical literature refers to quantitative financial alpha (cross-domain statistical dislocations and arbitrage indicators), not pre-release software status.
- An unauthenticated caller gets 30 calls a minute and 25 rows a response; a dollar of call credits, a pass or a plan raises both.
- Tool signatures and the CWF wire format are versioned: breaking changes ship behind a new major version.
- No uptime SLA is published, and none is implied. Call credits, passes, Builder and Team are list-priced and self-serve; Enterprise is a design partner engagement, because an SLA is what would make a list price at that level honest and there is not one yet.
- Prices are USD and settle in USDC. See ## Pricing & Plan Limits above or call `get_plans` for the current ladder.
## Core Capabilities
- **Model Context Protocol (MCP) Endpoint**: POST `https://pronto.stream/mcp` (or `/v1/mcp`)
- **15 Native MCP Tools**:
1. `get_billing_profile`: Report who you are resolved as and what your plan grants: the agent identity behind this call, the funded account it acts for, the plan, the per-minute call ceiling, the per-response row budget, and which indicator set you may compute.
2. `register_agent`: Create your own identity on pronto and receive an API key, with no human present.
3. `open_payment`: Buy access for yourself, with no human involved.
4. `settle_payment`: Redeem a payment you have made and activate what it bought — a plan, a pass or call credits.
5. `get_plans`: List what can be bought — call credits, day and week passes, and plans — what each costs, and what each raises — the call ceiling, the row budget, and whether the commercial indicator set is included.
6. `get_latest_signals`: Fetch the most recent live signals from the wire in token-efficient CWF format, optionally filtered to one collection.
7. `get_signal`: Fetch one signal by its id, when you already know which one you want.
8. `search_signals`: Find signals by the words in them.
9. `list_fusion_products`: List every derived intelligence product pronto can compute, with the published index or model each one implements, the signal collections it consumes, and whether it needs a plan.
10. `get_fused_signal`: Compute derived intelligence products from the live wire.
11. `scope_signals`: Select a slice of the live wire by facet tags — the structured metadata each signal carries, not the words in its text.
12. `get_facet_manifest`: Read the shape of the wire in one small response: a two-level tree of domain and the topics under it, each with a signal count, ranked so the busiest lead.
13. `list_facets`: List the full filter vocabulary: every facet value currently on the wire, across every key — lang, country, region, domain, topic, severity, provider, coverage and place — as one flat list with a signal count each, and a human-readable label for place ids.
14. `get_related_signals`: Follow the link graph from one signal to other signals about the same subject.
15. `list_data_sources`: List the data-source endpoints feeding the wire, with provider, signal type, country, cadence, and license.
## Live Data (cite freely, updated in real time)
- [Live Statistics Page](https://pronto.stream/seo/stats.html): Cited, dated metrics — lifetime signals processed, live country/domain/topic coverage, catalog reach, fusion products, CWF token savings.
- [Live Statistics JSON](https://pronto.stream/seo/stats.json): The same metrics as a schema.org Dataset with `variableMeasured` live values.
## Full Corpus & Discovery
- [llms-full.txt](https://pronto.stream/llms-full.txt): Expanded corpus — full tool surface, live metrics, plan limits, and live facet coverage in one fetch.
- [Agent/MCP discovery manifest](https://pronto.stream/.well-known/mcp.json): Auto-wire the MCP endpoint and related surfaces from one well-known URL.
## Profiles & Hosted Clients
- `https://pronto.stream/mcp` — the full surface. Use this unless your client needs one of the below.
- `https://pronto.stream/mcp/grok` — the Grok profile: the whole read surface plus `register_agent`, with the four commerce tools off the list and still callable. Paste it at grok.com/connectors under New Connector → Custom, or pass it as `server_url` to the xAI API.
- Codex: `codex mcp add pronto --url https://pronto.stream/mcp`. Optional credentials: `bearer_token_env_var = "PRONTO_API_KEY"` in the server configuration.
- `https://pronto.stream/mcp/openai` — read-only research profile with two tools, `search` and `fetch`, for clients that look them up by name.
- Any client that holds the MCP connection on its own infrastructure — Grok, ChatGPT, the Claude.ai connector — reaches pronto from a shared address, and every unregistered caller behind it shares one rate-limit window. Call `register_agent` once and send the key as an Authorization header: it costs nothing, needs no human, and gives you a window of your own.
## Data Sources & Licensing
- 99 verified structured API endpoints. Named examples, all live: USGS earthquakes, NOAA SWPC space weather, NWS alerts, Open-Meteo, MET Norway, World Bank WDI, US Treasury, ECB and Bank of Canada rates, SEC EDGAR, NIST NVD and CISA KEV, ClinicalTrials.gov, OpenFDA, OpenAlex, arXiv and Crossref, NASA DONKI and NeoWs. An endpoint counts here only once a live sweep has seen it return content.
- 96,814 RSS feeds across 230 countries, normalized in real time.
- Licensing varies by source and is reported per source by the `list_data_sources` tool. The wire as a whole is not under a blanket open licence: some sources are non-commercial or share-alike, and syndicated article text remains copyright its publisher. Aggregate terms: https://pronto.stream/terms
## SEO Topic Hubs & Guides
- [Knowledge Hub Index](https://pronto.stream/seo/): Complete developer documentation directory.
- [Best AI Data Labs & Alternative Data](https://pronto.stream/seo/best-ai-data-labs.html): Premier AI data labs and real-time alternative data providers for foundation models and post-training reasoning.
- [Alternative Data for AI Models](https://pronto.stream/seo/alternative-data-for-ai-training.html): Why frontier LLMs and quant funds require real-time empirical alternative data and 35 quantitative models.
- [MCP Server Guide for AI Agents](https://pronto.stream/seo/mcp-server-for-ai-agents.html): Setup for Codex, ChatGPT, OpenAI APIs and other MCP clients.
- [CWF Token-Saver Specification](https://pronto.stream/seo/cwf-token-saver-format.html): Technical spec for the compact wire format.
- [Fusion Products](https://pronto.stream/seo/fusion-products-guide.html): Cross-domain formulas and usage.
- [Signal Sources Guide](https://pronto.stream/seo/real-time-signal-sources.html): Ingestion pipeline and normalization engine.
- [Comparison](https://pronto.stream/seo/pronto-vs-polygon-finnhub-openmeteo.html): Feature matrix vs Polygon.io, Finnhub, and Open-Meteo.
- [Real-Time Alpha & Agent Arbitrage](https://pronto.stream/seo/real-time-alpha-and-agent-arbitrage.html): Cross-domain correlation vectors, token economics, and MCP-guided Base USDC settlement for quant agents.
- [Base AI Agents & On-Chain Infrastructure](https://pronto.stream/seo/base-ai-agents-and-onchain-signal-infrastructure.html): Autonomous sensory wire and MCP-guided Base USDC settlement for Base AI agents and Farcaster bots.
## Machine Readable Datasets & JSON Data
- [Evaluation Dataset Sample (CSV)](https://pronto.stream/pronto_signals_sample_1000.csv): Stratified 1,000-record production evaluation dataset under 14-field specification.
- [Commercial Data Dictionary](https://pronto.stream/pronto_data_dictionary.md): Specification v2.5.0 covering the 541M+ signal fabric.
- [Brand Facts JSON](https://pronto.stream/seo/brand-facts.json): Canonical facts, capabilities, and entity definitions.
- [Evidence Blocks JSON](https://pronto.stream/seo/evidence-blocks.json): Benchmark numbers, latency figures, and citable evidence.
- [OpenAPI Spec](https://pronto.stream/openapi.json): Full OpenAPI specification for HTTP REST routes.
- [Terms & Data Use](https://pronto.stream/terms): What may be reused, and why the answer is per source.
- [Sitemap](https://pronto.stream/sitemap.xml)
## Ecosystem & Partners
- [Ovasabi Studios](https://www.ovasabi.studio/): Parent AI, signal infrastructure, and systems engineering studio.
- [Vizi Financial Intelligence](https://vizi.ng/): Financial intelligence, payments, and real-time liquidity vector platform.