Kairos Signal MCP Server
Kairos Signal is an agent-facing MCP/data-commerce server for registering, managing credits, purchasing products, and querying/verifying DePIN and world datasets.
Register an agent with email/name to receive an API key and, per the schema, $5 starter credits (README says starter-credit eligibility is policy-dependent).
Browse purchasable products and prices, including DePIN telemetry, signal-feed tiers, DAG manifold API, GPU compute, and MCP unlimited access.
Purchase products with credits using an API key.
Top up credits via USDC transaction hash or Stripe checkout link, and check remaining balance.
List datasets and retrieve aggregate database statistics.
Fetch dataset records with limit/offset; free tier is limited to 10 records per query.
Verify dataset/record footprints with SHA-256 and retrieve cryptographic provenance for record IDs; README cautions these checks are not upstream-truth or zero-knowledge proofs.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Kairos Signal MCP Server Check my credit balance and pull the latest GPU supply snapshot for io.net and Akash."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Public proof of directed AI systems work — what I claim, what I refuse to claim, reproducible evidence: proof/
Kairos Signal MCP Server — DePIN and world data
Measured DePIN telemetry, physical-world datasets and descriptive research analytics through REST and MCP. Use source links, observation times and coverage metadata to decide whether a result is suitable for your question. Availability and freshness vary by feed.
Kairos Signal is not a property-owner contact or lead-generation service. The current product is data and research, not a promise of profitable trading signals.
Coverage changes. Use find_networks, get_data_dictionary and the network catalog rather than a copied network count. Catalog entries, distinct networks, Bittensor subnets, rows and metric series are different quantities. get_stats describes dataset row counts, not network coverage.
Connect directly
Full MCP endpoint (Streamable HTTP):
https://kairossignal.com/mcpRead-only edition, without account creation, purchases or payment tools:
https://kairossignal.com/mcp/publicRead-only does not mean every tool is anonymous or free. In either edition, get_network_signals requires an eligible paid-tier key with the engines capability. Other discovery/data tools have their stated no-key samples and limits. The full edition also exposes account tools and paid analytics.
Claude Code:
claude mcp add --transport http kairos-signal https://kairossignal.com/mcpCursor, Windsurf and clients that accept a URL:
{"mcpServers": {"kairos-signal": {"url": "https://kairossignal.com/mcp"}}}VS Code (.vscode/mcp.json):
{"servers": {"kairos-signal": {"type": "http", "url": "https://kairossignal.com/mcp"}}}Stdio clients can use this package's bridge:
{"mcpServers": {"kairos-signal": {"command": "npx", "args": ["-y", "kairos-mcp-server"]}}}Key placement depends on the tool:
get_network_signalsrequires an eligible subscription key in the connection'sX-API-Keyheader. Tiered network/history access also uses that header. In the stdio bridge,KAIROS_API_KEYsupplies the header.Paid analytics accept that header or an
api_keytool argument.purchase_data,topup_creditsandcheck_balancerequireapi_keyinside the tool input, even when the connection or bridge already has a header key. The bridge does not inject it into the arguments.
An API key, a credit balance and a subscription entitlement are not interchangeable.
Official MCP Registry identity: com.kairossignal/kairos-signal. Prefer the vendor endpoint above over a third-party proxy. Smithery's October 31, 2026 service notice says its proxy and connection-management services will close while its directory remains.
Related MCP server: JS-Peer x DeFi MCP Server
Ask a measured question
GPU capacity:
find_networks {"query":"akash"}, thenget_network {"symbol":"AKT","metric_filter":"gpu"}.Observed history:
get_metric_history {"symbol":"FIL","metric":"ACTIVE_MINERS"}. An absent series is not a zero.Check a figure:
verify_value {"symbol":"FIL","metric":"ACTIVE_MINERS"}. Read the verdict; some upstream values cannot be checked by a GET.
Tools and access
The bridge obtains tools/list from the hosted service rather than maintaining a separate implementation. Consult that live schema for current inputs, prices and limits.
Tool | Purpose | Access / effect |
| Scheduled Hodge/lag-relation snapshots for configured market, DePIN and world cohorts. Descriptive, not a trading recommendation. | Eligible paid-tier key with engines capability in the connection header |
| Find networks or subnets by name or ticker. Resolve ambiguous tickers before use. | No key |
| Latest supported first-party network measurements, with provenance and freshness fields. | No key |
| Observed daily history for one metric; gaps are not zero-filled. | No key for the free window; paid tiers have different limits |
| Compare with a supported upstream GET response, which may be cached; inspect | No key |
| Create an API key; starter credits are conditional on the current per-IP policy. | Creates an account/key |
| Product descriptions, prices and deliverability information. | No key |
| Purchase an eligible product; delivery can be inline or a receipt with a retrieval URL. |
|
| Request a supported card-checkout link or USDC quote; follow current payment instructions. |
|
| Read the account credit balance. |
|
| Current dataset catalog. | No key |
| Dataset row counts, scope and free-tier limits; not a count of distinct networks. | No key |
| Coverage, observation freshness, history depth and endpoints per feed. | No key |
| Index of covered archived fetches and their payload/code hashes; raw payload bytes are a separate product. | No key |
| Raw archive rows within the applicable query limits. Rejected or unchecked observations can be retained. | No key for the free sample |
| Merkle membership for a supported retained observation; not proof of upstream accuracy or zero-knowledge computation. | No key |
| Hodge decomposition of metric lead-lag flows; the historical tool name does not imply regime classification. | Paid per answered call |
| Correlation-spectrum comparison with a Marchenko–Pastur reference; not a calibrated regime test. | Paid per answered call |
| Enforcement-capacity calculation from sourced protocol parameters; bounds and unavailable inputs remain explicit. | Paid per answered call |
| Day/unit Shapley decomposition of observed metric changes; not causal attribution or a forecast. | Paid per answered call |
| Measured token-supply changes; not an exchange-listing or investment score. | Paid per answered call |
Paid analytics can spend credits. Read the current description before calling; do not repeatedly invoke a paid tool to retry an uncertain result. Their historically named identifiers are retained for compatibility, not as claims of forecasting or causal validation.
Provenance: what the checks establish
Supported network measurements expose provenance and observation-time fields. A source URL is a way to inspect an origin, not a guarantee that the origin is correct. History, derived analytics, raw archives and catalog metadata have different response contracts; do not assume that every output is a fresh first-party measurement.
verify_value compares against a supported upstream GET response that may be cached; inspect upstream.read_at_utc and the verdict rather than assuming a new upstream fetch. verify_footprint checks retained Merkle membership, not upstream truth, independent Bitcoin confirmation or a zero-knowledge proof. The derivation ledger covers the fetches it explicitly lists; a network can have live telemetry without an archived derivation entry. Raw archives can retain rejected or unchecked observations—read their quality fields.
Products, delivery and pricing
Use list_products and the live credit catalog for current prices, eligibility and deliverability; use the pricing page for subscription access. This documentation does not set prices or payment policy.
Registration creates a key; starter-credit eligibility depends on the current per-IP policy. Read the returned balance instead of assuming a grant. A starter key does not itself grant paid-tier engine access.
purchase_datarequiresapi_keyin its tool input and a currently deliverable product. Small responses can arrive inline; larger deliveries return a receipt and retrieval URL. Use your own stableidempotency_keyfor each intended purchase and reuse it only for retries of that purchase.The current catalog describes
dag_manifold_30das a DAG API access product. Listed legacy/retired products are not automatically purchasable; read deliverability before committing credits.topup_creditsprovides payment instructions for supported card and USDC modes. A quote or checkout link is not payment confirmation. Follow the payment documentation; never send funds without a successful current quote.
Links
Measurements and research only; not investment advice. Terms.
License: MIT
Available Tools
10 toolscheck_balanceAInspect
Check your remaining credit balance. Use after purchases to see remaining credits.
Inputs: api_key (required).
fetch_datasetAInspect
Query records from a dataset with limit/offset. Free tier: 10 records per query. Purchase products or top up credits for more.
Inputs: dataset (required), limit, offset.
get_zk_provenanceBInspect
Return cryptographic provenance proof for a record ID from current datasets.
Inputs: record_id (required).
list_datasetsAInspect
List all available datasets with record counts. Free to browse — market ticks, DePIN network stats, technical indicators, US county atlas, ZK footprints.
Inputs: no named parameters.
list_productsAInspect
Browse all purchasable products with prices in credits ($1 = 1 credit). Includes DePIN supply-telemetry data, signal feed tiers ($9.99-$399.90/mo), DAG manifold API ($2.99-$9.99/mo), GPU compute, and MCP unlimited access. Call after register_agent to see what to buy.
Inputs: no named parameters.
purchase_dataAInspect
Buy a product with your credits. Products: DePIN supply-telemetry snapshots, DAG manifold API (dag_pro/dag_quant/dag_enterprise), signal feed tiers, GPU compute, MCP unlimited. Use api_key from register_agent.
Inputs: api_key (required), product_key (required).
register_agentAInspect
START HERE. Self-register as an AI agent to get $5 free credits and an API key. No credit card, no human, no Stripe required. Credits work immediately for dataset queries, GPU inference, DAG manifold computation, and DePIN/API access. Call this first, then call list_products to see what you can buy.
Inputs: email (required), agent_name (required).
topup_creditsAInspect
Add credits when your $5 free balance runs out. Pay via USDC crypto (provide tx_hash) or get a Stripe checkout link for your human operator to complete.
Inputs: amount (required), method (required), api_key (required), tx_hash.
verify_footprintBInspect
Return SHA-256 cryptographic verification for a dataset or record
Inputs: dataset (required), record_id (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
Most tools have clearly distinct purposes: registration, product browsing/purchasing, balance, dataset listing, stats, fetching, and verification. The only mild overlap is between verify_footprint and get_zk_provenance, since both provide cryptographic assurances, but their descriptions are sufficiently distinct.
Tool names consistently use snake_case and follow a verb_noun pattern (register_agent, list_products, purchase_data, check_balance, list_datasets, fetch_dataset). Minor deviations like `topup_credits` as a single compound verb and `get_zk_provenance` with an acronym are still predictable and readable.
Ten tools is a well-scoped size for this server's mix of account management, marketplace purchasing, data access, and cryptographic verification. Each tool addresses a distinct stage of the user journey without unnecessary redundancy.
The tool surface covers the core lifecycle well: registration, products, purchase, balance management, dataset discovery, querying, and verification. A minor gap is the lack of explicit subscription management or finer-grained filtering options, but agents can otherwise complete the intended workflows without dead ends.
Maintenance
Related MCP Connectors
Curated marketplace of real-world data APIs for AI agents, paid per call in USDC on Solana.
Free market data with provenance and paid decision intelligence for AI agents.
Marketplace where AI agents buy datasets and API access, pay per call in USDC over x402.
Signed agent identity, trust scoring, credit economy, and social layer for AI agents.
Related MCP Servers
- -licenseNot gradedqualityNot gradedmaintenanceEnables AI agents to discover, purchase, and query blockchain-based datasets using Filecoin/IPFS storage with Ethereum NFT access controls and Lit Protocol encryption. Provides a decentralized marketplace for secure dataset trading with automatic purchase workflows and SQL query capabilities.0-
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to interact with peer-to-peer networks (libp2p) and DeFi protocols, including oracle networks, cross-chain communication, and intent-based execution.2MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to query cryptographically verified facts with zero-knowledge proofs, selective disclosure, and tamper-evident provenance.1Business Source 1.1

agentsoukofficial
AlicenseNot gradedqualityCmaintenanceEnables AI agents to create identities, list and find services, handle payments in USDC, and manage reputation through a decentralized marketplace.1MIT