Intelligence API & MCP, Documentation
The LyraMind Intelligence API turns market data, filings, and macro signals into grounded, cited, compliance-safe intelligence. 36 endpoints, 13 MCP tools. Every response carries a Trust Layer.
data_mode and never hides sample data.Quick start
# A grounded read on any symbol, no key needed
curl -X POST https://api.lyramindos.com/v1/company \
-H "Content-Type: application/json" \
-d '{"symbol":"NVDA"}'
# The go/no-go gate an agent checks before it acts
curl -X POST https://api.lyramindos.com/v1/readiness \
-H "x-api-key: $LYRAMIND_KEY" \
-d '{"symbol":"NVDA","action":"rebalance"}'
API host shown for illustration. The self-describing index lives at GET /v1; every response includes a trust{} envelope (confidence · sources · evidence · reasoning · data_mode · compliance).
Authentication & tiers
| Tier | Rate limit | How |
|---|---|---|
| anonymous | 60 / min (per IP) | no key, test freely |
| sandbox | 120 / min | built-in demo key lm_sandbox_demo |
| standard | 600 / min | x-api-key header |
| enterprise | unlimited | x-api-key header |
Responses set X-RateLimit-* and X-LyraMind-Tier headers. GET /v1/usage reports your own usage.
Endpoints
Market intelligence
| POST /v1/company | One-call company snapshot: read + confluence + risks + graph connections |
| POST /v1/explain | Plain-English "what & why" for a symbol (retail / pro) |
| POST /v1/committee | AI Investment Committee + LyraMind Score (5 lenses → composite) |
| GET /v1/pulse | Interpreted market-event feed over a universe |
| POST /v1/summarize | Condense text or a symbol's situation to key points |
Trust & verification
| POST /v1/readiness | Agent Readiness gate, AGENT_READY / REVIEW_REQUIRED / DO_NOT_EXECUTE |
| POST /v1/verify | Multi-reasoner verification, consensus, agreement, contradictions |
| GET /v1/memory | Institution Memory, a name's track record off the ledger |
| GET /v1/ledger · /verify · /stats | Trust Ledger, entries, hash-chain proof, calibration |
| POST /v1/ledger/outcome | Reconcile a past verdict (borne_out / contradicted / unclear) |
Graph · twins · agents · execution
| GET /v1/graph · /graph/cascade | Knowledge graph + macro→sector→company cascade |
| GET /v1/twin · POST /v1/twin/ask | Digital twin profile + conversational ask |
| GET /v1/marketplace · /run | Agent personas (investor / compliance / macro / catalyst) |
| POST /v1/execution/decide | Execution-Readiness, go/no-go before money moves |
| GET /v1/execution/providers | Pluggable execution-provider seam |
Corporate AI readiness · data
| POST /v1/ai-readiness/report | AI Readiness report for a draft filing |
| POST /v1/ai-readiness/simulate | Institutional AI Simulator |
| POST /v1/disclosure/scan | Disclosure legibility scan |
| GET /v1/ai-readiness/benchmark · /history | Peer benchmark + continuous monitoring |
| GET /v1/alpha/feed | Institutional Alpha Feed (gated) |
| GET /v1/questions/insights | Question Intelligence (Q-IQ) |
MCP server, 13 tools for agents
Expose the engine to any agent (Claude Desktop, agent frameworks, execution platforms) over Model Context Protocol. One-command install:
python services/mcp/setup.py # prints the config and verifies the server live
Tools: lyramind_agent_readiness · lyramind_execution_readiness · lyramind_record_outcome · lyramind_ledger_audit · lyramind_ask · lyramind_agent · lyramind_ai_readiness · lyramind_disclosure · lyramind_explain · lyramind_company · lyramind_committee · lyramind_pulse · lyramind_summarize. Every result carries the Trust Layer and a one-line agent-readable summary.