Send numeric scores to CI-1T and get back CI, EMA, authority levels, and ghost detection. Monitor model outputs, agent telemetry, or any confidence-like signal before failure looks normal.
Use the raw API if you want full control, the Python SDK if you want speed, or the MCP server if you want CI-1T available inside your tooling.
Post raw scores to CI-1T and get structured stability outputs back without running the engine or maintaining scoring infrastructure yourself.
Start with a single score stream, then grow into persistent sessions and multi-node fleet monitoring without changing the core contract.
Create an API key in the dashboard, fund credits, and move between REST, the Python SDK, and MCP tools from the same account.
Direct HTTPS endpoints for evaluate, fleet, sessions, lab, and probing workflows.
Typed Python client with auth, Q0.16 conversion, monitoring helpers, and fleet/session support.
Use CI-1T inside Copilot, Cursor, Claude Desktop, and other MCP clients without building your own UI first.
Monitor multiple nodes at once and compare stability across a live or historical window.
CI-1T accepts numeric score streams and returns stability metrics that are hard to derive reliably from a single confidence value. Use the same contract across models, agents, sensors, trading systems, or any other time-series of scores.
Use it for LLM output confidence today, then point the same API at sensor telemetry, ensemble outputs, or any other score stream tomorrow.
Start with the Python SDK for the fastest path to production, or call the API directly if you want CI-1T embedded deeper into your own services.
from ci1t import Client
client = Client("ci_...")
result = client.evaluate(scores=[0.72, 0.74, 0.12], n=3)
print(result["episodes"][-1]["ci_out"])Send scores in. Get stability metrics back. Keep your own alerting, policy, and product logic on your side of the boundary.
Pass a score stream and classify whether it is stable, drifting, flipping, or collapsing over time.
Track multiple nodes over time, compare windows, and inspect episode history across rounds.
Use CI, EMA, authority levels, and ghost flags to drive your own dashboards, alerts, or runtime policies.
pip install ci1t-sdk GitHub ↗ Live API keys, real credits, and production-facing integrations. The product is built to drop into real monitoring flows, not stay trapped in demos.
<200ms evaluations. Lightweight, zero runtime dependencies.
Used against live model traffic with inspectable run reports, session history, and drift analysis.
Sign in, create a CI-1T API key, fund credits, and call the engine directly from the API, SDK, or MCP server.
pip install ci1t-sdk. Three lines to your first evaluation. Auto Q0.16 conversion.
Create your API key in the dashboard, integrate through the API or SDK, and pay only for what you use. No monthly subscription treadmill. No enterprise sales detour.
Sign in, create an API key, and manage credits in one place before you ship an integration.
Use the engine when you need it. Pay for evaluations, not seats or annual contracts.
Use the raw API, the Python SDK, or MCP tools as the integration layer that makes sense for your product.