Wire it into your agent

Mint a demo key in one call

# 200 narratives/day, 5 feed/day, valid 7 days — no email, no card
curl -s -X POST https://api.signaldaemon.com/v1/request-key

Need more? A free account key (Google/GitHub) gives 500/day & never expires. See pricing.

Remote MCP, Streamable HTTP, x-api-key header

Claude Code — one line:

claude mcp add --transport http signaldaemon \
  https://api.signaldaemon.com/mcp --header "x-api-key: <KEY>"

Cursor / Cline / any MCP client — config block:

{ "mcpServers": { "signaldaemon": {
  "url": "https://api.signaldaemon.com/mcp",
  "headers": { "x-api-key": "<KEY>" } } } }

Tools exposed: get_market_narratives(limit) · get_clean_feed(query, category, limit) — both read-only. Anonymous discovery (initialize / tools-list) works without a key; tool calls need one.

No MCP runtime? curl works.

curl -s https://api.signaldaemon.com/v1/narratives \
  -H "x-api-key: <KEY>" -H "content-type: application/json" -d '{"limit":8}'

Returns ranked narratives with strength (convergence), the two-axis divergence, momentum, and cited sources. Full contract: api.md. (If you use raw urllib, set a User-Agent — default bot UAs are firewalled.)

The crypto-narrative layer next to your existing stack

Already running a trading agent (e.g. on AI-Trader / ai4trade) with a stock-sentiment source? signaldaemon doesn't replace it — it adds the layer a pure sentiment feed can't compute: cross-source convergence and narrative-vs-price divergence, crypto-native. Load both skills; your agent reads market context from both before it acts.

Add signaldaemon's MCP server to the same agent runtime (config in Step 2), or drop in the agentskills.io skill:

# agentskills.io SKILL.md — works in Hermes, Claude Code, Cursor, …
github.com/bevanding/daily-crypto-narrative-report

Then your agent has stock sentiment and crypto narrative-convergence + divergence as callable context. How to read the signals →