github Python analyzed 6858433

cognis-digital/evalbench

Offline LLM / agent eval harness with regression gates

Drift inferred · capture-to-capture

No drift recorded — single capability capture; advisories appear once its surface changes.

capabilities 1 tools

transport stdio counts 1 tools · 0 res · 0 prompts permission surface via code analysis

code evidence vHEAD · github-tarball

evidence-backed findings quoted directly from the published source artifact — not inferred

code files: 28

fs cognis-digital-evalbench-6858433/evalbench/cli.py :21 with open(path, "r", encoding="utf-8") as fh:
fs cognis-digital-evalbench-6858433/evalbench/connect.py :78 text = sys.stdin.read() if a.input == "-" else open(a.input, encoding="utf-8").read()
fs cognis-digital-evalbench-6858433/evalbench/core.py :459 with open(case["output_file"], "r", encoding="utf-8") as fh:
fs (weak) cognis-digital-evalbench-6858433/tests/test_evalbench_deep.py :236 base = evaluate_suite(json.load(open(BASELINE, encoding="utf-8"))).to_dict()
fs (weak) cognis-digital-evalbench-6858433/tests/test_smoke.py :66 with open(DEMO, encoding="utf-8") as fh:

net cognis-digital-evalbench-6858433/integrations/webhook.py :8 import argparse, json, sys, urllib.request
net cognis-digital-evalbench-6858433/livesearch.py :28 import urllib.parse