Conformal Prediction Python Packages

mapie

A scikit-learn-compatible library for estimating prediction intervals and controlling risks, based on conformal predictions.

crepes

Python package for conformal prediction

shad0wllm

System 0 for your LLM: learns the fixed-choice answers your LLM gives (intents, routing, moderation) and serves the repeats in microseconds from a small CPU table, at most 5% disagreement, certified. Zero-code proxy for the OpenAI Decisions API, System One and chat models.

distil-llm

Compression with a quality contract — cache-aware, causally-pruned LLM context compression for agentic runtimes, certified non-inferior across 7 domains. Works with any SDK.

solvi

A runtime where decision models propose and your code decides: typed questions, hard rules, quotes checked against the text, escalation with a guaranteed error budget, replayable traces. Works with any System One model.

system1

Not an LLM. Teach bounded decision skills from examples or observed APIs, validate, then run and save them locally. Optional MLX acceleration.

puncc

👋 Puncc is a python library for predictive uncertainty quantification using conformal prediction.

tsbootstrap

Fast, dependence-aware resampling and uncertainty for time series and ragged panels.

calibrated-explanations

Calibrated Explanations is an explanation method for uncertainty-aware explanations of classification and regression models.

loopbrake

Brakes for stuck AI agents. Learns how long your agent's tasks normally take, stops the one that runs way past that, and says why. For Claude Code, Codex and your own Python agents. Runs locally, zero dependencies.

undesirables-mcp-server

Zero-trust, local-first FastMCP core for The Undesirables. 34 offline AI agent tools: TCG card grading, conformal risk forecasts (Safe-Hold/Momentum grades), on-chain soul personalities, x402-payable oracle, TTS + 3D forging. No telemetry. BSL 1.1.

endgame-ml

A unified framework for tabular, time-series, and multimodal machine learning

labplan

Plan, fit, and certify laboratory calibrations for any user-supplied model: Fisher planning, D-optimal designs, conformal error bars, and audit-ready records. NumPy only.

vectrix

Zero-config time series forecasting for Python. 30+ models, auto selection, Forecast DNA, Rust turbo — one line of code.

sureband

Distribution-free coverage guarantees for System One decision models like Jev and Laya

cohera

Sheaf and hierarchy algorithms

certified-pruner

Stop bad runs early, with a conformal bound on how often you stop a good one

breezeml

A beginner-friendly, production-aware ML workflow layer for students, analysts, and AI agents. Train, compare, explain, export, and deploy scikit-learn models without the boilerplate.

noisefloor

Source-aware research diagnostics, correlation spectra, forecasting and synthetic Dyson experiments. Python, CLI, REST and MCP.

apdtflow

APDTFlow is a modern and extensible forecasting framework for time series data that leverages advanced techniques including neural ordinary differential equations (Neural ODEs), transformer-based components, and probabilistic modeling. Its modular design allows researchers and practitioners to experiment with multiple forecasting models and easily

grounded-reasoning

Zero-token, precision-guaranteed relation-algebra verifier for LLM/agent multi-hop reasoning

poorjev

Open-source, local Jev alternative: a System One decision layer with provably calibrated confidence (ECE 0.170→0.071). Typed decisions, runs offline, no API key, no waitlist.

certflow

Certified route planning under drifting costs: conformal certificates (LB<=OPT<=UB), certificate-directed sensing, proof-gated ns-us preprocessing, and an observable-coverage layer (WATCH + Shiryaev-Roberts). 250 tests, validated on real METR-LA.

poordjaevin

Djævin: calibrated local-first decision layer with a Devin ACP backend - no API key needed