Semgrep

Semgrep

@semgrep.com

Semgrep is a code scanning platform for finding first and third-party security vulnerabilities in your code base.

248 followers42 following503 posts

Will AI replace AppSec teams? 👀 In the latest episode of Security Rulez, our Security Advocate Dr. Katie Paxton-Fear (@InsiderPhD) sat down with Anshuman Bhartiya (Tech Lead at Lyft) to tackle this exact question. 👇

"AI agents are writing code so fast, we can't keep up with the security debt." 🫠 We have the solution: A plugin that lives in your IDE, detecting and resolving the vulnerabilities, malicious packages, and hardcoded secrets before a PR is ever opened.

mbt@1.2.48 and @cap-js/sqlite@2.2.2 are malicious. preinstall hook → fetches Bun runtime → executes obfuscated payload → exfils GitHub/npm/AWS/Azure/GCP/k8s secrets → writes to attacker-controlled GitHub repos with description "A Mini Shai-Hulud Has Appeared" we pushed rules for customers

PLEASE DON'T enable subcategory: - audit rules if you only want confirmed vulnerabilities, these are noisy by design! #EngineeringTakeover

This hack week some of us are building new features, others are writing talks, and some people are doing gods work. That big rock? Verified commits.

For years engineers wanted, and now they have it, everyone rejoices. Semgrep has a mobile rules editor #EngineeringTakeover

Rule writing: Whenever possible, avoid leading ellipsis patterns (These patterns cause the engine to perform exhaustive searches and are quite slow, we have rules for this) BAD ... <- VERY BAD DO NOT DO THIS $X = someFunc(...) ... GOOD someOtherFunc($X) ...

Here at Semgrep HQ our engineers are all in SF for our annual HackWeek so this is the Semgrep #EngineeringTakeover we've hacked the social accounts, we've locked out the marketing team and all posts going forward will be by engineers sorry, not sorry

Why are so many organizations still hesitant to truly experiment with AI security? Our Security Advocate, Dr. Katie Paxton-Fear, has the answer👇

Detect hard-coded JWT secrets in your Express.js codebase! Run: semgrep scan --config express-jwt-hardcoded-secret.yml ./src This rule catches risky credential patterns that could expose your authentication.

Want to better understand the Semgrep Multimodal approach? Rick Harp, Senior Solutions Engineer, explains what it is and how it’s different from other static analysis tools. 👇

Is your AppSec team scaling at the speed of AI, or are they still running on human-only hours? 🛡️ The timing is critical for two reasons.👇

AI-native assistants don't automatically understand your unique security practices during code development. Custom Guardrails bridge that gap.👇 #AppSec #SecureCode

If your team is leaning into "vibe coding" or heavy LLM usage, you need a strategy to ensure that speed doesn't turn into a liability. Here are four essential principles for securing AI-generated code👇

View full thread

4️⃣Use Offensive AI  AI isn't just the problem, it’s also the solution. Use "Offensive AI" to: - Detect emerging issues before they become exploits. - Automatically suggest fixes for vulnerabilities in your backlog. - Learn from past mistakes to harden your codebase over time.

If your team is leaning into "vibe coding" or heavy LLM usage, you need a strategy to ensure that speed doesn't turn into a liability. Here are four essential principles for securing AI-generated code👇

View full thread

2️⃣ Embed security early and continuously: Security can't be an afterthought. - In the IDE: Catch issues the moment they are written. - In the PR: Enforce scans automatically to ensure no vulnerable AI code ever reaches production.

3️⃣Enforce secure coding guardrails:  Think of guardrails as your security team’s "digital brain." By using custom rules and policies (like those in the Semgrep Pro Engine), you can set a definitive security posture that scales. If the AI suggests an insecure pattern, the guardrail stops it instantly.

If your team is leaning into "vibe coding" or heavy LLM usage, you need a strategy to ensure that speed doesn't turn into a liability. Here are four essential principles for securing AI-generated code👇

1️⃣ Assume AI code is vulnerable until proven safe: Our studies show that 48% of code generated by major LLMs contains vulnerabilities. You must apply the same (or even higher) scrutiny to AI suggestions as you would to a junior engineer's PR.

Want to scan your entire codebase without touching a single CI/CD file? 🛡️ In this quick walkthrough, we show you how to scale security across your repos in minutes using Semgrep Managed Scans. No manual config, just results.👇 #AppSec #SecureCode

Imagine an AI that reasons like a security engineer with the context of your lead developer.  Semgrep’s retrieval systems give any LLM the repo-specific nuance it needs to be reliable.👇 #AppSec

We provide secure coding feedback where it matters most: on the dev's screen. Faster feedback = less exploitable software, less frustration from devs and less time wasted.

If a vulnerable function in your supply chain isn’t reachable, it shouldn’t derail your sprint. If it *is* reachable, you need it at the top of the queue. Semgrep helps teams figure this out quickly so that remediation is efficient.

$ semgrep init --year 2026 [INFO] Initializing Future... [OK] [INFO] Deploying: Secure_Code.v2026 [SUCCESS] [WARN] Challenges: Loading...  Welcome to 2026❇️

Leave false positives in 2025. Imagine 2026: An AppSec world with zero noise and 100% developer trust. By leveraging the Semgrep platform, you can silence the friction of irrelevant alerts and focus on what actually matters ➡️ shipping secure code. 🌀Learn how we’re doing it: https://semgrep.dev/

59% of developers still don’t trust AI tools to handle security. With "vibe-coding" skyrocketing, even a small error rate creates a massive wave of new vulnerabilities. At Semgrep, we’re bridging that trust gap. #AppSec #AI #DevSecOps

In the world of "vibe coding," agents are powerful but they aren’t secure. Semgrep x Cursor Hooks changes that. Using Cursor Hooks allows AI agents to run and test code safely in their own environment, identifying vulnerabilities and applying fixes before you ever see the code.

That’s a wrap on Black Hat Europe 🇬🇧 Huge thank you to everyone who stopped by Booth #816 and to everyone who joined us at our events! We’re heading home feeling genuinely grateful for this community. Thanks for the great conversations, thoughtful questions, and good energy. Until next time! 👋

Huge thank you to everyone who joined us for Security Sundowners on the Sunborn Yacht last night 🛥️🍸 And a big shoutout to our partners who helped make it happen: Tines, Cyera, Sublime Security, and Zenity 🙌 #BlackHatEU #BHEU #AppSec #Cybersecurity #Semgrep

Still in town tonight? Join us for one more adventure: THE CUBE Experience – an AppSec Adventure 🧊 🕔 17:00–20:30 GMT | 📍 London (short tube from ExCeL) Teams of two, gameshow-style challenges + festive dinner & drinks. 👉 Register: semgrep.dev/events/the-c...

Black Hat Europe is in full swing, and we’re live at booth #816 with great conversations happening all day 🙌 Come say hi to the Semgrep team to see how our AI-driven AppSec platform helps dev and security teams fix vulnerabilities earlier, reduce noise, and accelerate release velocity.

The Semgrep team has touched down for Black Hat Europe! 🇬🇧 We’re set up and ready to see you tomorrow at Booth #816. Stop by to see how Semgrep’s AI-driven AppSec platform helps dev + security teams find and fix issues earlier, cut noise across SAST/SCA/Secrets, and ship faster.

Semgrep is wrapping up Black Hat EU with something a little different this year. Join us at THE CUBE live in London right after the conference for a gameshow-style AppSec adventure! 🗓️ Thu, Dec 11, 2025 | ⏰ 17:00–20:30 GMT 👉 semgrep.dev/events/the-c... We hope to see you there! 👋

It’s been a busy week for the Semgrep team! ⚡ 🎉 We’ve been out and about at AWS re:Invent, OWASP Benelux, the Colorado = Security Holiday Party, Merry & Mingling: Evolve's Holiday Mixer, and Hack the Halls! Curious where we’re heading next? Check out our events page: semgrep.dev/events/

Yesterday was a blast! 🏁🔥 A huge thank you to everyone who joined us at the Accelerate to Innovate – Raceday at re:Invent event with Sysdig. Amazing conversations, big energy, and such a fun mid-week highlight, we loved sharing it with you! Stop by booth #486 today before the week wraps! 🙌

What a night! 🌆✨ Huge thank you to everyone who joined us yesterday for Sip and Sync at the Sphere with Pellera Technologies and the AWS re:Invent Cocktail Reception with Felicis, Tableau, Supabase, Tines, and MotherDuck. Incredible conversations and such a fun way to kick off the week!

Tomorrow (Dec 4 @ 9:00am PT), join a fireside chat with Mudita Khurana (Airbnb) and Chushi Li (Semgrep) on measurement + benchmarks for the next generation of AppSec agents, including the Closed-Loop Capability (CLC) score. 🔗 Register here: semgrep.dev/events/measu...

Day 2 at AWS re:Invent is underway! If you're exploring the expo floor, come swing by Booth #486 to meet the team, check out live demos, and snag some exclusive Semgrep swag. #AWSreinvent #Semgrep #AppSec #DevSecOps #Cybersecurity

With all eyes on Sha1-Hulud, it’s easy to forget this is just the latest in a series of attacks showing a new normal where malicious dependencies wreak havoc on organizations. That’s why Malicious Dependency Detection is now generally available for Semgrep Supply Chain. semgrep.dev/blog/2025/bl...

The Semgrep team has officially landed in Las Vegas for AWS re:Invent! ✈️🎉 We’ll be at Booth #486 all week. Come meet the team, grab some great swag, and see how Semgrep helps engineering and security teams ship faster and stay secure. See you on the expo floor! 🙌

🇧🇪OWASP Benelux Days – see you tomorrow! 🇧🇪 We’re excited to touch down in Mechelen, Belgium soon – for a day of AppSec talks, followed by our ‘Brews & Bytes’ evening event afterwards… Learn more: semgrep.dev/events/owasp... Hope to see you there! #AppSec #InfoSec #Cybersecurity