
AgentCanary MCP
- Updated July 11, 2026
- MrCerq/agentcanary-mcp
AgentCanary is a MCP server that provides agents fresh market regime, risk, narrative, and scenario context via API key–authenticated tools.
About
AgentCanary is an MCP server (and REST API) that supplies AI agents with up-to-date market context: macro regime, risk posture, dominant narratives, and scenario state. Developers shipping financial copilots, risk-aware bots, or internal research agents can call standardized tools instead of stitching disparate macro feeds. Register for an AC_API_KEY at app.agentcanary.ai or via the documented wallet-based key creation URL, then run the agentcanary-mcp npm package over stdio in Claude Code, Cursor, or another MCP client. It sits in Build integrations when you code features, but the same data supports Idea research and Validate scoping when you sanity-check assumptions. The server follows the 2025-12-11 MCP schema and documents secrets handling for the API key environment variable.
- Market context API: regime, risk, narrative, and scenario state for agents
- Dual access via REST and MCP (npm agentcanary-mcp v1.4.4)
- Requires AC_API_KEY from app.agentcanary.ai or wallet-based key creation endpoint
- stdio MCP transport with documented environment variable
- Positioned as AgentCanary product with site at agentcanary.ai
AgentCanary MCP by the numbers
- Data as of Aug 10, 2026 (Skillselion catalog sync)
claude mcp add --env AC_API_KEY=YOUR_AC_API_KEY agentcanary-mcp -- npx -y agentcanary-mcpAdd your badge
Show developers this MCP server is listed on Skillselion. Paste this into your README.
| Package | agentcanary-mcp |
|---|---|
| Transport | STDIO |
| Auth | Required |
| Last updated | July 11, 2026 |
| Repository | MrCerq/agentcanary-mcp ↗ |
What it does
Wire fresh market regime, risk, narrative, and scenario state into agents building trading copilots, dashboards, or financial automations.
Who is it for?
Best when you're creating market-aware agents and want a keyed MCP integration plus REST parity from one vendor.
Skip if: Non-financial apps, developers who cannot use API keys, or teams needing exchange-level tick feeds only.
What you get
Your agent can query live regime, risk, narrative, and scenario state through MCP so decisions and code align with current market framing.
- Authenticated MCP tools returning regime, risk, narrative, and scenario snapshots
- Reusable market-context layer for financial agent features
- Documented secret env wiring for production-like local dev
By the numbers
- npm package agentcanary-mcp version 1.4.4
- Required env AC_API_KEY documented in MCP registry metadata
- Product site https://agentcanary.ai; GitHub repo MrCerq/agentcanary-mcp
README.md
AgentCanary MCP Server
MCP server for AgentCanary — the market context layer for autonomous AI agents. Fresh regime classifications, risk scores, narrative momentum, and scenario state via 19 MCP tools, with transparent source trails.
Quick Start
npx agentcanary-mcp
get_track_record is public and works without an API key. Set AC_API_KEY
for tiered AgentCanary API tools such as briefs, indicators, regime, signals,
and diagnostics.
Claude Desktop Config
{
"mcpServers": {
"agentcanary": {
"command": "npx",
"args": ["agentcanary-mcp"],
"env": { "AC_API_KEY": "ac_your_key_here" }
}
}
}
Cursor Config
~/.cursor/mcp.json:
{
"mcpServers": {
"agentcanary": {
"command": "npx",
"args": ["-y", "agentcanary-mcp"],
"env": { "AC_API_KEY": "ac_your_key_here" }
}
}
}
Restart Cursor → all 19 tools available via Composer.
Windsurf Config
~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"agentcanary": {
"command": "npx",
"args": ["-y", "agentcanary-mcp"],
"env": { "AC_API_KEY": "ac_your_key_here" }
}
}
}
Continue.dev Config
~/.continue/config.json → experimental.modelContextProtocolServers:
{
"experimental": {
"modelContextProtocolServers": [
{
"transport": {
"type": "stdio",
"command": "npx",
"args": ["-y", "agentcanary-mcp"],
"env": { "AC_API_KEY": "ac_your_key_here" }
}
}
]
}
}
ChatGPT (Custom GPT)
Don't want to install MCP? Use the AgentCanary Custom GPT directly in ChatGPT:
https://chatgpt.com/g/g-6a0f2b6ab07c819188983b3fb43ac62b-agentcanary
All major endpoints exposed as Actions. Free Explorer tier on the shared key.
Get an API Key
curl -X POST https://api.agentcanary.ai/api/keys/create \
-H "Content-Type: application/json" \
-d '{"walletAddress": "0xYourWallet"}'
Deposit USDC/USDT on any major EVM chain (Base, Ethereum, Arbitrum, Optimism, Polygon). $5 minimum. Credits never expire.
Tools (19)
| Tool | Tier | Returns |
|---|---|---|
get_briefs |
Explorer | Daily market intelligence briefs (4×/day: radar / signal / pulse / wrap) |
get_regime |
Builder | Current macro regime, risk gauge (0-100), key drivers, transition probabilities |
get_indicators |
Builder | Any of 36 proprietary indicators (Bull Market Support Band, Pi Cycle, Wyckoff, CAPE, Hindenburg…) — pass name= |
get_narratives |
Builder | Top active narratives with momentum scores, stage, asset impact |
get_predictions |
Builder | Prediction market data (Polymarket, Kalshi) |
get_scores |
Explorer | Prediction scoring results (hit / partial / miss after 72h) |
get_scenario_analysis |
Signal | Forward scenarios with price targets |
get_signals |
Builder+ | 20 sub-types via type=: whale-alerts, fear-greed, funding-rates, btc-etf-flows, vix, credit-stress, sector-rotation, insider-activity, correlations, dxy, oil, yield-curve, market-structure, stablecoin-dominance, whale-positions, cftc-cot, bofa-fms, dispersion, geopolitical-risk, decision-engine |
get_news |
Builder | Aggregated market news, filterable by ticker |
get_market_structure |
Builder | Orderbook depth, liquidation heatmap, exchange volumes — pass view= |
get_defi |
Builder | DeFi yields, stablecoins, chains, unlocks, perps — pass category= |
get_btc_options |
Builder | BTC options max pain + volatility skew |
get_central_banks |
Signal | Balance sheets, gold, reserves, TIC — pass view= |
get_expectations |
Signal | Market expectations (crowded, early, rotation) |
get_macro |
Builder | FRED, business cycle, global liquidity, M2, supply chain, high-impact calendar |
get_open_interest |
Builder | Cross-exchange OI across 43 perps + top by USD + 4h Δ% shifters |
get_liquidations |
Builder | 24h totals + 4h long/short split + per-side event counts + dominant-direction label |
diagnose |
Public | Reports whether a key is present; with AC_API_KEY, returns tier, scopes, credits, rate limit, and upgrade path. |
get_track_record |
Public | Mean Brier + per-asset hit rates + reliability buckets. No API key needed — same data as agentcanary.ai/record/. Optional ticker filter. |
Pricing
| Tier | Deposit | Calls/day | Adds |
|---|---|---|---|
| Explorer | free | 50 | briefs + scores |
| Builder | $50 USDC | 500 | + indicators / regime / narratives / news / predictions |
| Signal | $150 USDC | 2000 | + scenarios / positioning / full content |
| Institutional | $500 USDC | unlimited | white-label, SLA |
Per-call cost: $0.01-0.02 from deposit. Credits never expire. No subscriptions.
Links
License
MIT
Diagnose Tool
If any tiered tool returns tier_insufficient, scope_insufficient, or insufficient_credits, call diagnose for the exact upgrade path. Without AC_API_KEY, diagnose reports that only get_track_record is available publicly:
{
"mcp_version": "1.4.5",
"tier": "explorer",
"scopes": ["all"],
"credits": 50,
"upgrade_path": "Deposit $50 USDC on Base to upgrade to Builder tier"
}
Tier-Aware Errors
Errors are now structured. Examples:
[tier_insufficient]— endpoint requires a higher tier than your key carries[scope_insufficient]— your key has restricted scopes (re-issue with broader)[insufficient_credits]— top up via deposit on Base[rate_limited]— includesRetry after Ns
Recommended MCP Servers
How it compares
Keyed market-context MCP for agents, not a generic web search skill or portfolio execution terminal.
FAQ
Who is AgentCanary MCP for?
Developers building AI agents that need structured macro regime, risk, and narrative context with official MCP and REST access.
When should I use AgentCanary MCP?
Use it when coding or researching financial agents that must reason about current scenarios rather than static prompts.
How do I add AgentCanary MCP to my agent?
Obtain AC_API_KEY from https://app.agentcanary.ai or the documented create endpoint, install agentcanary-mcp, set AC_API_KEY in the MCP env block, and enable stdio transport.