AI-driven whitebox pentests, every finding reviewed by a certified engineer, from US$4,000 per test
Grade: C — Score: 60/100
Trace tests web apps and APIs the way an attacker would. In a whitebox test it clones the repositories you link into a temporary environment and reads the code; greybox and blackbox tests are offered when you cannot share source. Validation agents reproduce injection, XSS, access-control bypasses and SSRF against the environment you designate, preferably staging, and an OSCP or CISSP engineer reviews every finding. Each engagement ends with a report and a letter of attestation for SOC 2, ISO 27001 and HIPAA; PCI DSS and CASA coverage is on Scale and Enterprise.
Prices depend on billable lines of code and the number of applications. Startup costs US$4,000 per test for one web app or API up to 500,000 lines; Scale costs US$12,000 for up to five applications and adds mobile, desktop, CLIs and LLM surfaces such as MCP servers. Continuous coverage costs US$19,000 or US$48,000 a year, and Enterprise is quoted. The click-through terms set a 12-month subscription that renews automatically, so confirm how a one-time test is ordered.
All customer data is stored and processed in the United States, and testing uses Anthropic and OpenAI models. The privacy policy says customer data is not used for model training, while the standard terms Trace's contract incorporates allow aggregated, de-identified training. Documented connections are GitHub, GitLab, Bitbucket, AWS, Google Cloud, Vercel, Convex and Linear.
Startup (one-time engagement): US$4,000 per test
Scale (one-time engagement): US$12,000 per test
Startup (continuous coverage): US$19,000 per year
Scale (continuous coverage): US$48,000 per year
Enterprise: Custom quote
PR Reviewer (add-on): US$50 per developer per month; 50 PRs, then US$1 each
Consider switching to Traditional penetration testing firms: Trace says it is deeper than a manual pen test, delivered in days, and uses AI-native whitebox testing with real exploits and human review.
Trace says its pentest report and signed letter of attestation can be used for SOC 2, ISO 27001 and HIPAA, and that PCI DSS and CASA coverage is included only on the Scale and Enterprise tiers. An OSCP or CISSP engineer reviews every finding before the report is issued. Acceptance is still your auditor's decision, so confirm the scope, test dates and methodology they expect before you book the engagement.
Trace recommends a whitebox test, where it clones the repositories you link and tests the running application signed in and signed out. If you cannot share code, it also runs greybox tests with test accounts you provide and blackbox tests with no code and no accounts. The test type is agreed with the Trace team when the engagement is scoped, and Trace says whitebox finds the most because it can see why the code behaves as it does.
Trace clones your code into an isolated, temporary environment for the engagement and destroys it afterwards; what persists are the findings, the report and small code snippets at each affected location. Under its terms, customer data is kept while you have an active order and for 12 months after the last one, then deleted within 90 days, or within 60 days of a written request. Trace keeps a copy of each issued report and attestation letter for seven years. All customer data is stored and processed in the United States (AWS us-east-1).
Trace's privacy policy lists Anthropic and OpenAI as subprocessors for AI model inference and states that customer source code, findings and other customer data are not used to train models. However, its terms of service incorporate the Common Paper standard terms, whose machine learning section allows aggregated, de-identified customer content to be used to train or enhance AI models, and Trace's cover page does not exclude that section. If it matters to you, ask for a written no-training term in your Order Form. Trace says its model providers may keep requests for up to 30 days to detect abuse, and zero data retention is available on request.
No. AI agents analyze the code and attempt exploits at scale, but a certified security engineer reviews every finding before it reaches you, and Trace says it does not ship a finding it cannot substantiate. Engagements are also started by the Trace team only after the scope is agreed; nothing runs automatically.
Trace recommends a staging or pre-production environment, because its validation agents sign in with the test accounts you provide and try real exploits against the target. It says it does not exploit anything that could corrupt data, change persistent state or affect production; those findings are reported with code-level verification instead. Give it accounts in at least two tenants with more than one role so it can test whether one customer can reach another's data.
A scanner reports what might be vulnerable; Trace tries to prove impact. Its validation agents reproduce injection, XSS, access-control bypasses such as IDOR, and SSRF against your running application and record the request, session and data actually reached, and a security engineer signs off each finding. You also get a pentest report and letter of attestation, which a scan export does not provide.
Yes, on the Scale and Enterprise tiers, whose scope includes LLM surfaces such as MCP servers, AI chat, agents and the sandboxes they run in. Findings map to the OWASP Top 10 for LLMs as well as web, mobile and APIs, with examples such as prompt injection and excessive agency. The Startup tier covers one web app or API only.
Each engagement ends with a letter of attestation intended to be handed to auditors and reviewers, alongside the full pentest report. Trace's terms treat all findings, reports and testing results as your confidential information, so you decide what to share. The terms do not state who owns the report itself.
Check your Order Form. The pricing page sells Startup and Scale as one-time engagements priced per test, but Trace's click-through terms set a 12-month subscription period billed annually to the payment method on file, renewing unless you give notice at least 30 days before the period ends. Price changes apply at the next period after 30 days' notice. Confirm how your test is ordered and whether it renews before you sign.