Gawlo tests your LLM apps, AI agents, and MCP servers — plus the APIs and web behind them. It runs AI-specific probes (prompt-injection, tool poisoning, agent misuse).
A finding from a sent request carries that request and the response; a finding from static analysis is marked as such. It suggests a fix, and exports the results as an assessment report.
Gawlo is built for what modern apps actually ship — LLMs, agents, and MCP servers — and covers the classic surface too.
The classic attack surface behind your AI: zero-config API discovery, endpoint mapping, a Nuclei corpus of ~1,800 read-only misconfiguration, exposure, TLS and takeover checks, and an access-control tester for BOLA and IDOR. Scans are read-only by default; these checks write to your target and are opt-in: auth bypass, mass assignment, rate limiting, method-override probes, privilege-escalation RPC checks and mutating MCP tool calls.
A finding produced by a sent request carries that request and the response; a finding from static analysis (descriptor or configuration) is marked as such. Optionally, the model you configure explains the root cause and writes a code patch you can apply directly. Remediation is bring-your-own-key and stays off until you set a key.
Start free. Paid plans are coming. No hidden fees, no surprises.
For professional developers shipping production apps.
For teams that need shared visibility and controls.
For organizations with advanced compliance requirements.