Sentrint
a security scanner for projects build with LLMs
Last checked:
What is Sentrint?
Sentrint is a security scanning tool that analyzes code repositories built with LLM platforms (Claude, Gemini, ChatGPT, and 13 others) to identify vulnerabilities like hardcoded secrets, weak database access rules, vulnerable dependencies, and exploitable code paths. It uses AI filtering to reduce false positives and generates platform-specific fix prompts ready to paste back into the LLM that created the code. Users receive a security score (0-100), severity-categorized findings in plain English, and track improvements across multiple scans.
Sentrint pricing
Pricing model: Freemium
Free tier (no card required for ephemeral code scans); paid pricing not disclosed on this page
Sentrint pros
- Supports 14+ LLM platforms with tailored fix prompts for each platform
- AI-powered false positive filtering reduces noise and act-on findings
- Tests leaked credentials to verify which are actually exploitable/in-use
- Clear 0-100 scoring system with severity grouping, trend tracking, and before/after metrics on fixes
- Free tier available with no credit card required
Sentrint cons
- Only scans repositories you explicitly point it at; cannot discover or scan other repositories
- Cannot automatically rotate or reset leaked credentials—requires manual human action
- Focused specifically on LLM-built code; no mention of support for traditionally-written applications
Frequently asked questions about Sentrint
What does Sentrint scan for?
Hardcoded secrets (API keys, tokens, passwords), database access rules, open-source package vulnerabilities, code paths where user input runs as commands, and secrets in git history.
Which AI platforms does it support?
14 LLM platforms including Claude, Gemini, ChatGPT, Cursor, Lovable, Bolt, v0, and others. Fix prompts are rewritten for each platform.
How does it reduce false alarms?
An AI layer reviews every finding from the security engine and downgrades results that are not actually exploitable, filtering out noise.
How does the scoring work?
Sentrint assigns a grade out of 100, split separately for your code and dependencies, with findings grouped by severity (critical, high, medium, low).
Can Sentrint automatically fix vulnerabilities?
No—it generates a fix prompt you paste back into your LLM platform, but rotating leaked credentials and some manual steps require human action.