AZ-CLCE runtime

CLCE detects inconsistency, not intent. Type D is a label, not a finding of malice. Human validation required. Not a cybersecurity exploit, not a scanner of other people's systems, not a lie detector. Advisory scores only. Threshold 0.7 is the paper's acceptable line, not a pass/fail of truth.

OpenAPI: https://azclce-download-tracker.vibelock.workers.dev/openapi.json

Catalog: aziel-runtime.vibelock.workers.dev

curl -X POST https://azclce-download-tracker.vibelock.workers.dev/v1/score -H 'content-type: application/json' \
  -d '{"r":"login button blue","d":"login form submits","p":"login button submits"}'
curl -X POST https://azclce-download-tracker.vibelock.workers.dev/v1/classify -H 'content-type: application/json' \
  -d '{"r":"...","d":"...","p":"...","n":"csrf session"}'
curl -X POST https://azclce-download-tracker.vibelock.workers.dev/v1/gate -H 'content-type: application/json' \
  -d '{"r":"a","d":"a","p":"a","min":0.7}'

GET/POST under /v1 never increment the download counter.

Downloads