s64-orbital-v1 / open_source /hello_aicoevolution.py
AICoevolution
Paper 03 research bundle (HF) 2026-02-02 10-04
1ceda33
#!/usr/bin/env python3
"""
hello_aicoevolution.py
======================
Minimal working example for the PyPI package:
pip install aicoevolution
Usage (recommended):
# Windows PowerShell:
# $env:AIC_SDK_API_KEY="aic_..."
# cmd.exe:
# set AIC_SDK_API_KEY=aic_...
python hello_aicoevolution.py
Optional:
set AIC_SDK_URL=https://sdk.aicoevolution.com
"""
from __future__ import annotations
import os
import time
import uuid
# This script is intended to be run on any machine after:
# pip install aicoevolution
#
# Note: If you run this file *inside the MirrorMind monorepo*, Python may import
# the local `aicoevolution_sdk/` package instead of the PyPI thin client module.
# For end users, copy this file into a normal folder (or a fresh venv) and run it there.
from aicoevolution_sdk import AICoevolutionClient, SDKAuth
def _print_metrics(label: str, resp: dict) -> None:
"""Print SDK metrics with Paper 03 canonical extraction.
Paper 03 metrics:
- SGI: turn_pair_sgi_latest (or fallback to sgi_latest)
- Velocity: orbital_velocity_latest (turn-pair, ~25-45°)
Fallback: angular_velocity_latest (per-message, ~75-180°)
The SDK now exposes these at top level for easy access.
"""
print(f"\n[{label}]")
print("conversation_id:", resp.get("conversation_id"))
print("message_count:", resp.get("message_count"))
# Paper 03: Prefer turn-pair SGI when available (now at top level)
sgi = resp.get("turn_pair_sgi_latest") or resp.get("sgi_latest")
# Paper 03: Prefer orbital velocity (turn-pair, ~25-45°) over angular (per-message, ~75-180°)
velocity = resp.get("orbital_velocity_latest") or resp.get("angular_velocity_latest")
# Core telemetry fields (may be None early in a conversation)
print("sgi_latest:", sgi)
print("orbital_velocity_latest:", velocity)
print("context_phase:", resp.get("context_phase"))
print("context_mass:", resp.get("context_mass"))
print("attractor_count:", resp.get("attractor_count"))
quota = resp.get("quota")
if isinstance(quota, dict):
remaining_units = quota.get("remaining")
try:
remaining_units_int = int(remaining_units)
turns_left = max(0, remaining_units_int // 2) # ~2 ingest calls per Q/A turn
print("turns_left_this_week:", turns_left)
except Exception:
# If quota is unlimited (-1) or missing, just print the raw object.
print("quota:", quota)
def main() -> None:
api_key = (os.getenv("AIC_SDK_API_KEY") or "").strip()
if not api_key:
raise SystemExit("Missing AIC_SDK_API_KEY. Set it to your aic_... key and retry.")
base_url = (os.getenv("AIC_SDK_URL") or "https://sdk.aicoevolution.com").strip()
sdk = AICoevolutionClient(
base_url=base_url,
auth=SDKAuth(user_api_key=api_key),
timeout_s=30.0,
)
conversation_id = f"hello_{uuid.uuid4().hex[:8]}"
now_ms = lambda: int(time.time() * 1000)
# 1) Ingest user message
r1 = sdk.ingest(
conversation_id=conversation_id,
role="user",
text="Hello — I'm testing AICoevolution semantic telemetry.",
timestamp_ms=now_ms(),
)
_print_metrics("USER INGEST", r1)
# 2) Ingest assistant message (this is when SGI/velocity typically becomes meaningful)
r2 = sdk.ingest(
conversation_id=conversation_id,
role="assistant",
text="Great. Tell me what topic you'd like to explore and I’ll respond concisely.",
timestamp_ms=now_ms(),
)
_print_metrics("ASSISTANT INGEST", r2)
if __name__ == "__main__":
main()