2 min read
Add as a preferred source on Google

Detectify’s Agentic Tooling Tests Suspected Flaws in Running Apps

Customer-run AI agents handle one reported issue per ticket, while a Security Ledger keeps an add-only record of test requests and evidence.

A single application test runs beside a locked evidence record / TokenPost.ai
A single application test runs beside a locked evidence record / TokenPost.ai

Detectify’s Agentic Tooling lets organizations use AI agents to examine individual suspected security issues in live applications and preserve evidence from each test. The Swedish security company’s workflow ties an agent’s determination to a specific reported concern.

A case can begin with a finding from static application security testing (SAST), a bug bounty report or a self-reported issue from an internal red team. Each ticket covers one suspected issue. An agent can inspect relevant code in the customer’s environment and test the application, then return one of four outcomes: confirmed, not confirmed, inconclusive or not testable.

Detectify’s Security Ledger keeps an add-only record of tool activity, test requests and results. The agent submits its determination, while the tooling writes the supporting evidence into the ledger. Agents cannot edit that stored record, which preserves the evidence alongside the verdict.

The agent runs in the customer’s environment, where it can inspect code and reach the application under test. Detectify does not access or store customer codebases. The information it receives includes investigation data, ledger entries, ticket contents and request metadata. SAST exports can include code snippets.

Potentially harmful checks, including denial-of-service tests, are marked unsafe so a person can decide whether to proceed. That review step applies to tests that could disrupt the application being examined.

Detectify announced its MCP Server on May 26, 2026, to connect vulnerability findings and attack-surface data with AI workflows. Agentic Tooling includes testing tools, a ticket queue and an evidence record. An earlier story about AI agents for enterprise vulnerability work covered investigation and fix verification; Detectify’s described process focuses on testing a specific suspected issue in an application.

Simon Yoon

Reporter

Simon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.

Loading…