Mantis documentation

Everything you need to install, run, configure, and extend Mantis. Pick a starting point below or jump straight to the quickstart.

Start here

Clone, install, pick a harness, run your first scan. About 60 seconds end-to-end.

Claude, GPT-5, Gemini, DeepSeek, Llama, anything on OpenRouter or local Ollama. Per-agent model overrides.

The 7-phase FSM, MCP control plane, specialist agents, safety rails. Diagram and walkthrough.

DISCOVER, REASON, TEST, LEARN. How a praying mantis hunt maps onto offensive-security automation.

RECON → AUTH → HUNT → CHAIN → VERIFY → GRADE → REPORT. Every transition, every legal back-edge.

All 12 agents: orchestrator, recon, triage, hunter, chain-builder, three verifiers, grader, reporter, patcher, discloser.

The brutalist → balanced → final verification contract. Why a finding needs three re-proofs to ship.

API reference for every typed tool: mantis_http_scan, mantis_record_finding, the full wave + verification + grade family.

fast (pre-screen) / standard / ultra (high-throughput) / loop (long missions) / fullsend (auto-disclose). Cost vs. validity matrix.

Cloudflare, Akamai, AWS WAF, GCP Armor entries. Auto-selected from the surface's detected tech stack.

scope-guard (out-of-scope blocker), session-write-guard, self-defense rules, deny-list, manual review gate.

Resume an interrupted run, grade a HOLD finding, auto-signup against a target, single-shot a known surface.

Tip

If you're brand new, start with the Quickstart. Want to use OpenAI / Gemini / Llama? Jump to Use any AI model.