- Category
- Cybersecurity
- Rank
- No. 3893Tools index
Previous survey · No. 3678 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Security & Identity
- GitHub
- 5 stars
- Latest release
- v0.4.2
- Date
About
Rowan is a static security scanner for AI applications. It reads source code, model files, dependencies and agent/MCP configuration without running the code, and reports likely issues such as injection, unsafe deserialization, SSRF, leaked secrets, risky agent tools and unsafe model loading. Each finding comes with evidence to review, and the project is marked alpha, so findings are leads to verify rather than confirmed bugs.
At a glance
Rowan scans source and model files for likely security issues without executing the project. It is worth trying as an initial review if you want specific lines to inspect for risky tool calls, model loading or exposed credentials.
Good fit
Developers triaging a codebase they are authorized to scan and able to review each finding.
Consider something else when
Skip it as a release sign-off or if you need to demonstrate that an issue is exploitable at runtime.
Start here
Read the documented self-test, then plan a scan of a synthetic project with one known unsafe pattern. Check whether the finding explains the relevant line and what you would need to confirm.
Cost and requirements
MIT-licensed alpha project. The quickstart requires Python 3.10 or newer and the Opengrep engine.
Limits to know
A reported issue is a lead for manual review. A clean result leaves open risks outside the configured rules and patterns that static analysis misses.
Related work
- Bouncer Security Check (complement)
Sources checked 2026-11-05. Source-based assessment.
Intel on Rowan
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.