Skip to content

// PUBLIC BENCHMARKS

Honest lab metrics — with caveats

We publish reproducible harness results: an in-process regression fixture (BENCH-01), multi-lab probes (BENCH-02), and novel-target runs (GEN-01b). Numbers below are from real runs — not marketing targets.

Full methodology & raw JSON on docs →

BENCH-01 — in-process regression fixture (3 recall + 2 precision cases)

In-process regression fixture (3 recall + 2 precision cases). Starlette micro-lab — measures engine regression, not full-app recall on customer targets.

3/3
Recall (vulns found)
2/2
Precision (safe endpoints)
0
False positives (CONFIRMED)
CasePhaseResultNotes
recall_cmdi_pingcmdiPASSshell via host= (audit deep test)
recall_xss_searchxss_wafPASSunescaped HTML reflection — WAF bypass phase
recall_ssti_greetsstiPASStemplate eval {{7777*7777}}
precision_safe_cmdicmdiPASShtml.escape — must not CONFIRM CmdI
precision_safe_xss_wafxss_wafPASShtml.escape — ENGINE-18 / audit H1

Novel targets (GEN-01b) — clean motor, no lab seeds

Autonomous surface-driven scans on apps we do not tune against. Only published after a real run.

TargetRecall (core)Recall (extended)CoverageNotes
OWASP NodeGoat3/38/9100%playwright_crawl OK (13 pages, 6 DOM forms) · 6 injection point(s) on bus · discover_api: 0 API endpoint(s) · sources={} · recall_core 3/3 on NodeGoat truth (`dast_detectable: yes`) — surface-driven motor (GEN-02): SSJI eval timing, profile write XSS, open redirect · no seed pack · truth matching is keyword/CWE heuristic

Multi-lab probes (BENCH-02)

Targeted recall probes on known labs (Jul 2026). Skipped labs are not counted as failures. Full scorecard: docs.

2/2 labs passed · 1 skipped (lab not reachable)

LabResultHitsNotes
Broken CrystalsPASS1/1GraphQL introspection / misconfig — SPA demo target
OWASP Juice ShopSKIPPED0/1REST search XSS recall on public demo instance
VAmPI (local)PASS1/1BOLA on /users/v1/{username} — docker compose in demo/

Last run: 7/2/2026, 7:50:04 PM