See it run.

from git push to signed verdict · 17s loop

pull request opened

acme-ml/zerogpu-demo · #142open · ci queued

feat: add inference latency cacheacme-dev wants to merge 3 commits into main from feat/inference-cache

● LookingEye verdict · queuedtriggered by .lookingeye.yml

signing with sigstore

lookingeye-report.jsonv1 · building

{
  "schema": "lookingeye.report/v1",
  "space": "acme-ml/zerogpu-demo",
  "pr": 142,
  "decision": {
    "status": "advisory",
    "warning_codes": [
      "ZEROGPU_WORKER_FAIL",
      "QUOTA_BURN_NO_OUTPUT"
    ]
  },
  "manifest_sha256": "9c7e2a…b13f"
}

sigstore · cert chain ok · tlog #78340912

signed

POST lookingeye.draftlabs.org/api/cli/ping

202 · accepted

live demo · auto-play01 · pr opened

A developer pushes a PR. The repo already has .lookingeye.yml committed, so GitHub Actions queues the LookingEye workflow automatically.

what you sawLookingEye running inside GitHub Actions on a sample PR

what it caughtTwo substrate failure modes HF would have hidden behind a 200 OK

what shippedA Sigstore-signed verdict bundle, pinned to the PR commit

ready to emit

Run this flow on your own repo.

The Free CLI ships everything you just watched. Add the hosted PR-bot when you want verdict history and the advisory→required gate.

Sign up to get the free CLIFree forever · no card required