Security Scan Report
Automated multi-tool security scanning (SCA, SAST, IaC, secrets detection) of this JavaScript/Node.js project produced 518 findings across six tools. No P0-level indicators were detected. However, synthesis elevated rollup (CVE-2026-27606) from High to Critical based on NVD CVSS 9.8.
AI analysts independently discovered 14 additional high-severity application security vulnerabilities that no scanner caught — including an IDOR, path traversals, stored XSS, and SQL injection.
| # | Severity | Package & Title | CVE | CVSS |
|---|---|---|---|---|
| 1 | Critical | [email protected] — Arbitrary File Write via Path Traversal | CVE-2026-27606 | 9.8 |
| 2 | High | [email protected] — DoS via Uncontrolled Recursion | CVE-2026-3520 | 8.7 |
| 3 | High | [email protected] — ReDoS via Repeated Wildcards | CVE-2026-26996 | 8.7 |
| 4 | High | [email protected] — DoS via Resource Exhaustion | CVE-2026-2359 | 8.7 |
| 5 | High | [email protected] — Code Injection via _.template | CVE-2026-4800 | 8.1 |
| 6 | High | @remix-run/[email protected] — XSS via Open Redirects | CVE-2026-22029 | 8.0 |
| 7 | High | [email protected] — WebSocket Unhandled Exception | CVE-2026-2229 | 7.5 |
| 8 | High | [email protected] — WebSocket 64-bit Length Overflow | CVE-2026-1528 | 7.5 |
| 9 | High | [email protected] — Unbounded Memory in WebSocket | CVE-2026-1526 | 7.5 |
| 10 | High | [email protected] — Ed25519 Signature Forgery | CVE-2026-33895 | 7.5 |
8 AI security analysts examined the codebase for threats that traditional tools can't detect.
startsWith() which can be bypassed. Allows reading arbitrary files on the server; admin users can write to arbitrary paths via .builder/../../.env.new RegExp(field.pattern) constructs a regex from user input with no complexity bound. Attacker-crafted patterns cause catastrophic backtracking, blocking the event loop.| Package | Previous | Fixed | Severity | CVEs |
|---|---|---|---|---|
| rollup | 4.46.2 | 4.59.0 | Critical 9.8 | CVE-2026-27606 |
| multer | 2.0.2 | 2.1.1 | High 8.7 | CVE-2026-3520, CVE-2026-2359 |
| minimatch | 9.0.5 | 9.0.7 | High 8.7 | CVE-2026-26996, CVE-2026-27903, CVE-2026-27904 |
| hono | 4.12.1 | 4.12.4 | High 8.2 | GHSA-xh87-mx6m-69f3, GHSA-q5qw-h33p-qvwr |
| lodash | 4.17.21 | 4.18.0 | High 8.1 | CVE-2026-4800 |
| @remix-run/router | 1.23.0 | 1.23.2 | High 8.0 | CVE-2026-22029 |
| undici | 7.23.0 | 7.24.0 | High 7.5 | CVE-2026-2229, CVE-2026-1528, CVE-2026-1526 |
| node-forge | 1.3.3 | 1.4.0 | High 7.5 | CVE-2026-33895, CVE-2026-33894 |
| path-to-regexp | 8.2.0 / 8.3.0 | 8.4.0 | High 7.5 | CVE-2026-4926 |
| picomatch | 2.3.1 / 3.0.1 / 4.0.3 | 4.0.4 | High 7.5 | CVE-2026-33671 |
| fast-xml-parser | 5.5.3 | 5.5.6 | High 7.5 | CVE-2026-33036 |
| glob | 10.4.5 | 10.5.0 | High 7.5 | CVE-2025-64756 |
| @anthropic-ai/sdk | 0.80.0 | 0.81.0 | Medium 6.3 | CVE-2026-34451 |
Based on this scan, Thresher generated a remediation PR that resolved 2 Critical, 28 High, and 5 Medium findings — including 13 dependency upgrades, 12 application security fixes, 1 secret rotation, and 6 CI/CD hardening measures.
Adversarial verification: An independent agent challenges all AI findings, filtering false positives and validating exploit paths.
Synthesis: Cross-track agreement analysis with priority elevation/downgrade logic. Scanner and AI findings are merged, deduplicated, and ranked by composite severity.