fahali 0.1.0

Market intelligence AI agents can verify - signed receipts, judged track record (misses included), Brier-calibrated forecasts. Adapters for LangChain, CrewAI, LlamaIndex, AutoGen, smolagents. MCP: mcp.fahaliai.com

fahali

Market intelligence AI agents can verify. Every verdict carries a signed SHA-256 receipt. Every judged signal has a public replay URL. The track record keeps its misses and publishes its methodology. Forecast probabilities are Brier-scored against realized outcomes.

Zero dependencies. Works with any agent framework. Observation, not advice.

pip install fahali    # Python
npm install fahali    # JS/TS

Free key (50 calls/day, no card): app.fahaliai.com/developer · Docs: fahaliai.com/developer · MCP: https://mcp.fahaliai.com/mcp

Direct

import { FahaliClient } from "fahali";

const fahali = new FahaliClient({ apiKey: process.env.FAHALI_API_KEY });
const verdict = await fahali.verdict(["BTCUSDT"]);   // committee read + signed receipt
const record  = await fahali.symbolRecord("BTCUSDT"); // judged record, misses included

OpenAI (and anything using function-calling format)

import { FahaliClient, FAHALI_TOOLS, executeTool } from "fahali";

const fahali = new FahaliClient({ apiKey: process.env.FAHALI_API_KEY });
const tools = FAHALI_TOOLS.map((t) => ({ type: "function", function: t }));
// ...pass `tools` to chat.completions.create; on a tool call:
const result = await executeTool(fahali, call.function.name, JSON.parse(call.function.arguments));

LangChain (Python)

from fahali import FahaliClient
from fahali.langchain import get_langchain_tools
tools = get_langchain_tools(FahaliClient(api_key=...))

CrewAI

from fahali.crewai import get_crewai_tools

LlamaIndex

from fahali.llamaindex import get_llamaindex_tools

AutoGen / smolagents / Semantic Kernel / raw OpenAI

from fahali import FahaliClient, TOOL_SPECS, execute_tool  # function-calling format

Vercel AI SDK

import { tool, jsonSchema, generateText } from "ai";
import { FahaliClient, toAiSdkTools } from "fahali";

const fahali = new FahaliClient({ apiKey: process.env.FAHALI_API_KEY });
const { text } = await generateText({
  model,
  tools: toAiSdkTools(fahali, { tool, jsonSchema }),
  prompt: "Is anything threatening a BTC-heavy portfolio right now? Cite replays.",
});

LangChain.js

import { DynamicStructuredTool } from "@langchain/core/tools";
import { FahaliClient, toLangchainTools } from "fahali";

const fahali = new FahaliClient({ apiKey: process.env.FAHALI_API_KEY });
const tools = toLangchainTools(fahali, { DynamicStructuredTool });

Tools

Tool	What your agent gets
fahali_verdict	18-engine committee read: agreement, dissent on the record, signed receipt, replay URL
fahali_forecast_72h	Crash/neutral/pump distribution, Brier-calibrated
fahali_tape	Latest judged calls market-wide — hits and misses
fahali_symbol_record	Per-symbol judged record with replay URLs
fahali_track_record	Aggregate scorecard: lift vs base rate, sample sizes, disclosed gaps
fahali_recent_alerts	Live detections across ~600 scanned instruments

Why this instead of a terminal or search

Terminals were built for human eyes; search gives agents unaccountable text. Fahali's answers are claims your agent can check: methodology at fahaliai.com/methodology, pricing free → metered at fahaliai.com/developer.

MIT © Future Legends AI. Nothing here is a recommendation to buy or sell anything.