Trickest

Trickest operates a security execution platform that turns offensive security testing into composable workflow graphs. Users build workflows on a visual canvas as directed acyclic graphs of pre-integrated security tools, then execute them at scale on managed or self-hosted machines; individual nodes can be burst across a fleet and re-executed with upstream dependencies memoized. Tool output is normalized into typed "Live Tables" that sync with run data, are queryable through the Trickest Query Language and a write API, and can be surfaced in dashboards.

The platform is accessible through a web builder, a CLI, a typed TypeScript SDK (@trickest/sdk), an API, and an MCP integration for coding agents such as Claude Code, Cursor and Codex. In July 2026 the company released Agents, which build workflow graphs from a described task, place agent nodes beside tool nodes, and run each agent in its own isolated sandbox for tools, files, and runs. Product areas marketed to security teams include offensive security (discovery, testing, validation, red team as continuous workflows) and exposure management (finding reachable assets, validating exploitability, and tracking the attack surface). Additional capabilities shipped over time include Modules (packaged, auto-updating tools), Private Agents for scanning internal networks, private tooling and a private Docker registry, self-hosted execution machines, advanced variables, RBAC, SAML SSO, and audit logs.

Founding story

Founded in 2020 by Nenad Zarić and Mihailo Tomić. Zarić was a security professional and bug bounty hunter who held top rankings on HackerOne for programs including Uber, PayPal and Snapchat, and has described the field's shortage of practitioners as forcing individuals to do the work of several people — the problem the platform was built to address. The team describes itself as engineers, former bug bounty hunters and technology people focused on making security testing composable and reproducible.

Business model

Software platform sold on a tiered model: a free Community edition for a single user with three self-hosted execution slots, the workflow builder and public workflow library, an AI agent using the customer's own model, CLI/SDK/API/MCP access and 7-day data retention; and a custom-priced Teams & Enterprise tier adding managed and self-hosted execution at scale, multiple users and shared workspaces, modules and private tooling, RBAC, SAML SSO, audit logs, flexible hosting, onboarding, training and dedicated support. The company has publicly argued against per-asset pricing in attack surface management in favor of pricing execution capacity.

Free Community tier for individual developers and custom-priced enterprise subscriptions covering platform access and execution capacity, presented as a single API and a single bill.

Traction

Case studies cite a financial technology customer reducing adversary simulation from three weeks to six hours per tactic and a fast-moving consumer goods customer scaling from eight to thirty security tests per month with the same team. The company reported $3M in total funding as of mid-2024 and maintains a regularly updated product and research blog through 2026.

Latest developments

In July 2026 Trickest introduced Agents, which compose real tools into a DAG from a described task and execute across the fleet alongside handwritten workflows, with each agent isolated in its own sandbox. Recent blog output covers deterministic workflows as a response to AI-accelerated attacks, slopsquatting of AI-hallucinated packages, security team burnout and automation, continuous validation versus point-in-time testing under DORA, NIS2 and SEC requirements, and consolidation of offensive security stacks.