Noya Agent

Noya is a multi-agent AI system for crypto trading, prediction markets (Polymarket, Rain), token analysis, and DCA strategies, plus a large suite of public structured-data endpoints (prices, TVL, funding rates, liquidations, wallet analytics, news, sentiment, prediction-market intelligence, AI-scored token catalog). All on-chain transactions are gas-sponsored — users pay no gas fees.

• Website: agent.noya.ai
• Agent API Base URL: https://agent-api.noya.ai (requires NOYA_API_KEY)
• Data API Base URL: https://data-endpoints.noya.ai (no API key, no account needed)
• Docs Base URL: https://mcp.noya.ai (public, no auth — /llms.txt for index, /llms.mdx/docs/{path}/content.md for full page content)

Trust & Security

• All API calls use HTTPS. Only NOYA_API_KEY is read from the environment, and it is only needed for the conversational agent endpoints.
• The public data endpoints at data-endpoints.noya.ai are unauthenticated and can be called without any setup.
• All on-chain transactions require explicit user confirmation via an interrupt prompt before execution.
• Use a short-lived API key (30-day) for testing. Revoke it from Settings > API Keys if compromised.

Setup

The public data endpoints (see Data Endpoints) work immediately with no setup.

To additionally use the conversational agent endpoints (messaging, threads, chat completions, user summary, agent summary):

1. Create an account at agent.noya.ai
2. Go to Settings > API Keys and generate a key
3. Store the key securely — it is only shown once
4. Set the environment variable:

export NOYA_API_KEY="noya_your_key_here"

Configure in ~/.openclaw/openclaw.json:

{
  "skills": {
    "entries": {
      "noya-agent": {
        "enabled": true,
        "apiKey": "noya_your_key_here",
        "env": {
          "NOYA_API_KEY": "noya_your_key_here"
        }
      }
    }
  }
}

When to Use

Use Noya when users want to:

• Check token prices or portfolio balances
• Swap, bridge, or send tokens (cross-chain supported)
• Analyze tokens and market trends
• Trade on Polymarket or Rain prediction markets
• Set up or manage DCA (dollar-cost averaging) strategies
• View DeFi positions
• Start a voice conversation with the Noya agent
• Pull deterministic crypto data: spot/historical prices (CoinGecko), TVL & yields (DeFiLlama), funding rates / open interest / liquidations (CoinGlass), wallet balances / transfers / DeFi & NFT positions (Moralis), crypto news with sentiment (CryptoNews), Fear & Greed index, on-chain DEX pools & trades (GeckoTerminal), AI-scored token catalog, Polymarket market intelligence, Kaito social intelligence (mindshare, volume-weighted sentiment, smart-follower graph, catalyst events)

Routing guidance:

• Use the data endpoints (no API key, direct curl) for deterministic lookups — prices, TVL, funding rates, news, on-chain analytics, prediction-market discovery, Kaito mindshare/sentiment/smart-follower data. Faster and cheaper than going through the agent graph.
• Use the agent (noya-message.sh) for anything that requires reasoning, execution, or the user's connected wallet: swaps, bridges, transfers, DCA setup, placing Polymarket orders, personalized portfolio Q&A.

Don't use for: Non-crypto tasks, local file operations, or general knowledge questions.

Core Workflow

Every interaction uses the Noya REST API. The primary endpoint is POST /api/messages/stream which returns a streamed response. Always use the provided noya-message.sh script to send messages — it handles streaming, chunk parsing, and formatted output. Do not call the message endpoint with raw curl.

OpenClaw integration: For every new chat OpenClaw initiates with Noya, first call the system message endpoint (step 2.5) to hand off conversation context. This makes the chat feel like a seamless continuation of the user's conversation with OpenClaw.

0. Read the Docs (when unsure)

When you need to understand a tool's parameters, response shape, or usage patterns, fetch the docs directly as plain markdown — no API key required:

# Full index of all doc pages
curl -s https://mcp.noya.ai/llms.txt

# Full content of a specific page
curl -s https://mcp.noya.ai/llms.mdx/docs/data-endpoints/coinglass/content.md
curl -s https://mcp.noya.ai/llms.mdx/docs/data-endpoints/moralis/content.md
curl -s https://mcp.noya.ai/llms.mdx/docs/data-endpoints/coingecko/content.md

The index lists every page with its description. Use it to discover what's available, then fetch the specific page for parameter details and examples. Always check the docs before guessing at a response shape.

1. Discover Capabilities (first use)

curl -s -H "x-api-key: $NOYA_API_KEY" \
  "https://agent-api.noya.ai/api/agents/summarize" | jq '.data'

Returns all available agent types (token analysis, prediction markets, DCA, etc.) and their tools. Call this once to understand what Noya can do.

2. Generate a Thread ID (new conversation)

Generate a UUID v4 for each new conversation topic:

python3 -c "import uuid; print(uuid.uuid4())"

Or on macOS/Linux:

uuidgen | tr '[:upper:]' '[:lower:]'

Each conversation needs a unique UUID. Generate one per topic and reuse it for follow-ups.

2.5 Initialize Thread with Conversation Context (OpenClaw)

For every new chat that OpenClaw initiates with Noya, call this endpoint first to set up the conversation context. This makes the chat with Noya feel like a natural continuation of the user's conversation with OpenClaw, rather than starting fresh.

curl -s -X POST "https://agent-api.noya.ai/api/openclaw/system-message" \
  -H "Content-Type: application/json" \
  -H "x-api-key: $NOYA_API_KEY" \
  -d '{
    "threadId": "THREAD_ID_HERE",
    "content": "The user has been chatting with OpenClaw and now wants help with crypto/trading tasks. Here is the relevant context from our conversation:\n\n- The user mentioned they have a meeting at 3pm today and want to check their portfolio before then\n- They previously expressed preference for ETH over BTC\n- Their timezone is EST\n- Earlier in our chat, they asked about setting up a DCA strategy\n\nPlease continue assisting them naturally, as if this is a seamless handoff from our conversation."
  }'

The content field should frame the handoff as a conversation continuation. Include:

• A brief intro explaining the user was chatting with OpenClaw and is now being handed off to Noya
• Relevant context from the OpenClaw conversation (what the user asked about, their goals)
• User's schedule, preferences, and any other helpful details
• A note to continue the conversation naturally

This ensures the user doesn't have to repeat themselves and Noya can pick up where OpenClaw left off.

Response:

{
  "success": true,
  "filtered": false,
  "message": "Content was sanitized before appending"
}

The content passes through a security filter. If filtered is true, the content was sanitized before being added. If the content is rejected entirely, you'll receive a 400 error with a reason field.

Important: Call this endpoint before sending the first user message via noya-message.sh. The system message will be prepended to the thread's context.

3. Send Messages (streaming)

Use the provided script to send a message and receive the parsed response:

bash {baseDir}/noya-message.sh "What tokens do I have in my portfolio?" "THREAD_ID_HERE"

The script handles the streaming response, parses --breakpoint-- delimited JSON chunks, and outputs formatted text including messages, tool results, progress indicators, and interrupt prompts.

4. Continue the Conversation

Reuse the same thread ID for follow-ups — Noya maintains context:

bash {baseDir}/noya-message.sh "Swap 0.1 ETH to USDC" "SAME_THREAD_ID"

5. Respond to Interrupts

When the agent needs confirmation (e.g., before executing a swap), the output includes [REQUIRES INPUT] with options. Send the user's answer as a follow-up in the same thread:

bash {baseDir}/noya-message.sh "Yes" "SAME_THREAD_ID"

Start Voice Chat

Opens the Noya AI agent in voice chat mode in the user's browser. Use this when the user wants to talk to Noya by voice instead of text.

Always include threadIdToUse when OpenClaw initiates the voice chat. Since step 2.5 (set system message) creates the thread beforehand with conversation context, you must pass that thread ID to continue the same conversation:

open "https://agent.noya.ai?mode=voice&threadIdToUse=THREAD_ID_HERE"

Only omit threadIdToUse if the user explicitly asks to start a completely fresh voice session without any context:

open "https://agent.noya.ai?mode=voice"

On Linux, use xdg-open instead of open.

API Reference (curl commands)

All endpoints require the x-api-key header. Base URL: https://agent-api.noya.ai

Send Message (streaming) — ALWAYS use the script

Do not call /api/messages/stream with raw curl. The response is a custom streamed format that requires parsing. Always use the provided script:

bash {baseDir}/noya-message.sh "<message>" "<threadId>"

The script handles authentication, streaming, --breakpoint-- chunk parsing, and outputs clean formatted text (messages, tool results, interrupts, progress, errors).

List Threads

curl -s -H "x-api-key: $NOYA_API_KEY" \
  "https://agent-api.noya.ai/api/threads" | jq '.data.threads'

Get Thread Messages

curl -s -H "x-api-key: $NOYA_API_KEY" \
  "https://agent-api.noya.ai/api/threads/THREAD_ID/messages" | jq '.data.messages'

Delete Thread

curl -s -X DELETE -H "x-api-key: $NOYA_API_KEY" \
  "https://agent-api.noya.ai/api/threads/THREAD_ID"

Get Agent Summary

curl -s -H "x-api-key: $NOYA_API_KEY" \
  "https://agent-api.noya.ai/api/agents/summarize" | jq '.data'

Get User Summary (all holdings, DCA strategies, Polymarket positions)

Returns a single structured snapshot of everything relevant to the authenticated user — ideal for feeding as context to another agent.

curl -s -H "x-api-key: $NOYA_API_KEY" \
  "https://agent-api.noya.ai/api/user/summary" | jq '.data'

Response includes:

• holdings — all wallet tokens and DeFi app positions with USD values
• dcaStrategies — all DCA strategies (active, inactive, errored, completed)
• polymarket.openPositions — current open prediction market positions with PnL
• polymarket.closedPositions — 20 most recently closed positions

Set System Message (OpenClaw)

Injects a system message into a thread before the conversation starts. OpenClaw should call this for every new chat to hand off conversation context to Noya, making the transition feel seamless for the user.

curl -s -X POST "https://agent-api.noya.ai/api/openclaw/system-message" \
  -H "Content-Type: application/json" \
  -H "x-api-key: $NOYA_API_KEY" \
  -d '{
    "threadId": "THREAD_ID_HERE",
    "content": "The user has been chatting with OpenClaw and now wants help with crypto tasks. Context from our conversation: ... Please continue assisting them naturally."
  }'

Request body:

• threadId (string, required) — The thread ID to attach the system message to
• content (string, required) — Conversation context framed as a handoff from OpenClaw to Noya

Response:

• success (boolean) — Whether the operation succeeded
• filtered (boolean) — Whether the content was sanitized by the security filter
• message (string, optional) — Present if content was sanitized

Errors:

• 400 — Invalid request or content rejected by security filter (includes reason)
• 401 — Unauthorized (invalid API key)

Data Endpoints (no API key)

Noya also exposes a large set of public, unauthenticated structured-data endpoints at https://data-endpoints.noya.ai. These are ideal for deterministic lookups — they do not route through the agent graph, so they are faster and require no NOYA_API_KEY.

Conventions:

• All POST endpoints take a JSON body with the parameters below. GET endpoints take none.
• All endpoints return JSON from the upstream provider.
• No auth header required.

Prefer these over noya-message.sh whenever the user just wants raw data.

alternative.me

GeckoTerminal (on-chain DEX data)

CoinGecko

Note: CoinGecko endpoints use token IDs (e.g. bitcoin, ethereum), not symbols. Use /coingecko/search first if you only know the symbol.

CoinGlass (derivatives)

CryptoNews

DeFiLlama

Moralis (wallet analytics)

chain accepts names (eth, bsc, polygon, arbitrum, base, optimism, avalanche) or hex IDs. toBlock is supported only on /moralis/wallet (historical token-balance snapshot); the DeFi-positions and NFT-holdings endpoints return current state only.

Noya Tokens (AI-scored catalog)

Noya Polymarket (prediction-market intelligence)

Kaito (social intelligence)

Kaito requires entity resolution first: before any analytics call that takes token/tokens, resolve the identifier via /kaito/entities. Narrative IDs are case-sensitive — resolve via /kaito/narratives.

Kaito interpretation rules (apply across tools):

• sentiment_score is volume-weighted and absolute — compare against the entity's own historical range, not fixed thresholds. sentiment_score is NOT smart_engagement.
• Mindshare rank: Top 10 dominant, Top 20 strong, Top 50 moderate, >50 weak. If all-zero data is returned for a ticker, retry with the full project name (e.g. HYPE → HYPERLIQUID).
• Narrative movement classification: Surging ≥ +10%, Fading ≤ −10%, Stable within ±10%.
• Spike detection for engagement/mentions: flag any day exceeding 2× the period average; cross-reference with /kaito/events.
• Smart follower tiers: 5,000+ Elite, 1,000–5,000 Strong, 300–1,000 Solid, 50–300 Emerging, <50 Low.
• /kaito/events returns forward-looking catalysts only — always cite the per-event references URL. Flag events within 7 days as imminent.
• Advanced-search rules: put time in min_created_at/max_created_at, not in query. Prefer tokens alone when the project resolves; Kaito ANDs all fields so every added parameter narrows results. Twitter-only sort options (smart_engagement, engagement, sentiment, bookmark, views, author) aren't compatible with News — use relevance or created_at across both.

Batch (parallel multi-request)

Dispatch up to 20 data endpoints in a single HTTP call. Each sub-request runs in parallel through the same rate limiter and cache as a direct call, so cache entries are shared and partial failures are isolated (one failing item never fails the whole batch).

Each requests[] item:

• name (string, required, ≤100 chars) — caller-supplied unique key; results are returned as results[name]
• method (string, required) — GET or POST
• path (string, required) — full endpoint path starting with /, e.g. /coingecko/price. Nested /batch calls are rejected.
• body (object, optional) — JSON body for POST endpoints
• query (object, optional) — query params for GET endpoints that take input via query string
• timeoutMs (integer, optional) — per-item timeout override in ms, capped at 60000

Top-level defaultTimeoutMs (integer, optional) sets the default per-item timeout (default 15000, max 60000). Batch size is capped at 20 items; names must be unique within a batch.

curl -s -X POST "https://data-endpoints.noya.ai/batch" \
  -H "Content-Type: application/json" \
  -d '{
    "requests": [
      { "name": "btc-price",   "method": "POST", "path": "/coingecko/price",
        "body": { "tokenIds": ["bitcoin"], "vsCurrencies": ["usd"] } },
      { "name": "fear-greed",  "method": "POST", "path": "/alternative/fear-greed",
        "body": { "limit": 1 } },
      { "name": "btc-funding", "method": "POST", "path": "/coinglass/funding-rates",
        "body": { "symbol": "BTC" } }
    ]
  }' | jq

Response shape:

{
  "results": {
    "btc-price":   { "ok": true,  "status": 200, "data": [...], "cache": "HIT" },
    "fear-greed":  { "ok": true,  "status": 200, "data": {...}, "cache": "MISS" },
    "btc-funding": { "ok": false, "status": 504, "error": "Sub-request timed out after 15000ms" }
  }
}

Each result is either { ok: true, status, data, cache? } or { ok: false, status, error }. cache is "HIT" or "MISS" when the sub-request went through the cache layer. Use /batch whenever you'd otherwise fire multiple sequential curls to the same base URL — it reduces round-trips without changing per-item semantics.

Calling a data endpoint with curl

# POST with a JSON body (no auth header)
curl -s -X POST "https://data-endpoints.noya.ai/coingecko/price" \
  -H "Content-Type: application/json" \
  -d '{"tokenIds":["bitcoin","ethereum"],"vsCurrencies":["usd"],"include24hrChange":true}' \
  | jq

# GET (no body, no auth)
curl -s "https://data-endpoints.noya.ai/coingecko/trending" | jq

# Fear & Greed
curl -s -X POST "https://data-endpoints.noya.ai/alternative/fear-greed" \
  -H "Content-Type: application/json" -d '{"limit":7}' | jq

If the user provides a plain symbol ("SOL", "ARB"), resolve it to a CoinGecko ID first:

curl -s -X POST "https://data-endpoints.noya.ai/coingecko/search" \
  -H "Content-Type: application/json" -d '{"query":"sol"}' | jq '.coins[0].id'

Common Patterns

Check Portfolio

User: "What's in my wallet?"

1. Generate a thread ID: uuidgen | tr '[:upper:]' '[:lower:]'
2. bash {baseDir}/noya-message.sh "What tokens do I have in my portfolio?" "$THREAD_ID"
→ Returns wallet address, token balances, and portfolio data

Token Swap

User: "Swap 0.5 ETH to USDC"

1. Generate a thread ID
2. bash {baseDir}/noya-message.sh "Swap 0.5 ETH to USDC" "$THREAD_ID"
→ Noya prepares the swap, asks for confirmation (interrupt), then executes.
   All gas fees are sponsored. User must confirm before execution.
3. bash {baseDir}/noya-message.sh "Yes" "$THREAD_ID"  # after user confirms

Token Analysis

User: "Analyze SOL for me"

1. Generate a thread ID
2. bash {baseDir}/noya-message.sh "Give me a detailed analysis of SOL" "$THREAD_ID"
→ Returns price data, market trends, and analysis

DCA Strategy

User: "Set up a DCA for ETH"

1. Generate a thread ID
2. bash {baseDir}/noya-message.sh "Set up a weekly DCA strategy for ETH with $50" "$THREAD_ID"
→ Noya configures the DCA strategy and confirms details

Prediction Markets

User: "What are the top Polymarket events?"

1. Generate a thread ID
2. bash {baseDir}/noya-message.sh "Show me the top trending Polymarket events" "$THREAD_ID"
→ Returns current events, markets, and trading options

Full User Context for Another Agent

Use case: You need to brief another AI agent on everything about the user
before delegating a task to it.

1. curl -s -H "x-api-key: $NOYA_API_KEY" \
     "https://agent-api.noya.ai/api/user/summary" | jq '.data' > user_context.json
2. Pass user_context.json as the system/user context to the downstream agent.
→ Returns wallet holdings, all DCA strategies, open and closed Polymarket
  positions in a single JSON object. Partial failures are isolated — the
  response always returns whatever data is available, with an error field
  for any source that failed.

Quick Price Check (no API key)

User: "What's the price of BTC and ETH right now?"

curl -s -X POST "https://data-endpoints.noya.ai/coingecko/price" \
  -H "Content-Type: application/json" \
  -d '{"tokenIds":["bitcoin","ethereum"],"vsCurrencies":["usd"],"include24hrChange":true}' | jq

→ Returns USD prices and 24h change. No agent round-trip, no thread ID.

Market Sentiment Snapshot (no API key)

User: "How's market sentiment today?"

1. curl -s -X POST "https://data-endpoints.noya.ai/alternative/fear-greed" \
     -H "Content-Type: application/json" -d '{"limit":1}' | jq
2. curl -s "https://data-endpoints.noya.ai/cryptonews/trending?date=last7days" | jq
→ Fear & Greed index + trending tickers from crypto news.

Wallet Snapshot for Any Address (no API key)

User: "What's in wallet 0xabc...?"

curl -s -X POST "https://data-endpoints.noya.ai/moralis/wallet" \
  -H "Content-Type: application/json" \
  -d '{"address":"0xabc...","chain":"eth"}' | jq

→ Works for any address. For the user's own connected wallet, prefer
  /api/user/summary (agent API) which also includes DCA & Polymarket.

Discover Top Polymarket Opportunities (no API key)

User: "Show me the best Polymarket bets right now"

curl -s -X POST "https://data-endpoints.noya.ai/noya/polymarket/top-ev" \
  -H "Content-Type: application/json" -d '{"limit":10}' | jq

→ Use for research. To actually place a trade, switch to the agent
  via noya-message.sh (requires API key + user wallet).

Voice Chat

User: "I want to talk to Noya"

1. Generate a thread ID
2. Call set system message endpoint with conversation context (step 2.5)
3. open "https://agent.noya.ai?mode=voice&threadIdToUse=$THREAD_ID"
→ Opens voice chat with OpenClaw's conversation context already loaded

User: "Start a fresh voice chat with Noya, no context needed"

1. open "https://agent.noya.ai?mode=voice"
→ Opens voice chat without any prior context (only when user explicitly requests)

Important Notes

Transaction Confirmation

Noya always asks for user confirmation before executing on-chain transactions (swaps, bridges, transfers, orders). The response will include a [REQUIRES INPUT] section with details and options. Always relay this to the user and send their answer as a follow-up in the same thread. Never auto-confirm transactions.

Wallet Delegation (Website Only)

If Noya responds with a delegation request, the user must complete this on the website:

"To delegate your wallet, visit https://agent.noya.ai and click
'Delegate Wallet' in the chat. This is a one-time action."

Safe Deployment (Website Only)

If Noya responds with a Safe deployment request, the user must complete this on the website:

"To deploy your Polymarket Safe, visit https://agent.noya.ai and click
'Deploy Safe Now'. This is free, takes ~30 seconds, and only needs to be done once."

Error Handling

Agent API (agent-api.noya.ai):

Data API (data-endpoints.noya.ai):

Scripts

This skill includes the following script in its directory:

Additional Resources

• For the complete REST API specification, see {baseDir}/reference.md