PyPI page
Home page
Author:
flobi
License:
MIT
Summary:
Cognitive observability for LLM agents. @styxx.profile returns a per-step cognometric readout: drift, confabulation, refusal, sycophancy, phase transition, low trust, incoherence. Register instruments with documented ceilings: guardrail.check vs a grounding passage, HaluEval-QA AUC 0.998 (fails DROP, FinanceBench); refuse_check, XSTest-v2 0.976; drift_check, BFCL v3 0.943. Reference implementation of Cognometric Fingerprint Spec v1.0. No torch, no GPU, no LLM in the loop; base install: numpy, scikit-learn.
Latest version:
7.49.0
Required dependencies:
numpy
|
scikit-learn
Optional dependencies:
anthropic
|
circuit-tracer
|
crewai
|
cryptography
|
guardrails-ai
|
jsonschema
|
langchain-core
|
langfuse
|
langsmith
|
llama-index-core
|
matplotlib
|
mcp
|
openai
|
pillow
|
psutil
|
pyautogen
|
pytest
|
ruff
|
scikit-learn
|
scipy
|
sentence-transformers
|
torch
|
transformer-lens
|
transformers
Downloads last day:
40
Downloads last week:
217
Downloads last month:
575
7-day smoothing