Scan levels

One entry for the whole product. Give a URL, a connected repo, a lockfile or a package list, pick how deep to go, and get one report with a grade per level.

The levels

LevelCovers
L1 SurfaceSecurity headers, CSP and cookies, CORS, robots.txt AI-crawler rules, llms.txt / ai.txt, bot-protection signals.
L2 ExposureEverything in L1, plus exposed .env, .git and source maps, secrets in JavaScript bundles, public OpenAPI / GraphQL / API docs, prompt-injection surfaces. The default.
L3 Supply chainEverything in L2, plus the Eridas OSV scan: packages the site's bundles announce with a version, or a connected repo, lockfile or package list.
L4 ReconstructionEverything in L3, plus REA: what an attacker's agent can rebuild from your public bundles and source maps. Local only; shown as unavailable on hosted deployments.
Device (coming soon)Scan the machine that builds and runs the app.

Each level includes every level below it. Levels that were not run are listed as not run, never silently passed.

Run it

# CLI
npx eridas tinfoil https://example.com --level 2
npx eridas tinfoil --level 3 --lockfile package-lock.json
npx eridas tinfoil --level 3 --packages lodash@4.17.15,minimist@1.2.0

# Cloud API (erd_ key or signed-in session)
curl -X POST $ERIDAS_API_URL/api/v1/tinfoil \
  -H "Authorization: Bearer $ERIDAS_API_KEY" -H "Content-Type: application/json" \
  -d '{"url":"https://example.com","level":3}'

# SDK
await client.tinfoilScan({ url: "https://example.com", level: 3 });

MCP exposes eridas_tinfoil_scan (local) and eridas_cloud_tinfoil_scan (Cloud), both with a level argument.

Honest limits

Level 3 from a URL only sees packages that announce their version in public bundles. For a complete answer add a lockfile or connect the repo. Scans are passive and read-only: GET requests, no credentials, and private, loopback and metadata addresses are refused.