Detectify launches Agentic Tooling and Security Ledger to verify AI-agent vulnerability findings in live applications
Detectify has launched Agentic Tooling, a product news item rather than a threat or incident. Customers' own AI agents can use it to test whether suspected vulnerabilities are exploitable in running applications. The tools run in the customer's environment through MCP and a CLI, with the customer's own agents and models. A Security Ledger records each testing session, including requests, payloads and results. The ledger is written by Detectify's tools rather than the agent, and it is append-only. This lets teams tell tested findings from unsupported AI-generated conclusions.
Detalhes técnicos
- The tools draw on Detectify's testing engines, payloads from its ethical hacker community, Alfred AI and Surface Monitoring data.
- The product targets "loop engineering": a CI run or scheduled job turns findings into tickets, and an agent tests each one against the live application.
- It builds on the Detectify MCP Server released in May 2026.
- The article cites Detectify's claim that about 80% of AI-driven SAST findings are noise. It also cites a Fudan University and Tencent study that found 76% of static-analysis alerts were false positives.
- Detectify's H2 2026 Cyber Hygiene Index found that nine in ten open critical and high-severity vulnerabilities stayed exposed for more than 90 days.
The article is based on a vendor announcement. No vulnerabilities, threat actors or IOCs are involved, and no external research was done.
Linha do tempo
-
Detectify launches Agentic Tooling and Security Ledger to verify AI-agent vulnerability findings in live applications
ler a análise desta fonte ocultar
Detectify has launched Agentic Tooling, a product news item rather than a threat or incident. Customers' own AI agents can use it to test whether suspected vulnerabilities are exploitable in running applications. The tools run in the customer's environment through MCP and a CLI, with the customer's own agents and models. A Security Ledger records each testing session, including requests, payloads and results. The ledger is written by Detectify's tools rather than the agent, and it is append-only. This lets teams tell tested findings from unsupported AI-generated conclusions.
Technical Details
- The tools draw on Detectify's testing engines, payloads from its ethical hacker community, Alfred AI and Surface Monitoring data.
- The product targets "loop engineering": a CI run or scheduled job turns findings into tickets, and an agent tests each one against the live application.
- It builds on the Detectify MCP Server released in May 2026.
- The article cites Detectify's claim that about 80% of AI-driven SAST findings are noise. It also cites a Fudan University and Tencent study that found 76% of static-analysis alerts were false positives.
- Detectify's H2 2026 Cyber Hygiene Index found that nine in ten open critical and high-severity vulnerabilities stayed exposed for more than 90 days.
The article is based on a vendor announcement. No vulnerabilities, threat actors or IOCs are involved, and no external research was done.