Dele (@DeleLearnsTech) header
Dele (@DeleLearnsTech) profile picture

- Platform Security Engineer - University Lecturer - Medium profile : https://teapawt.org/t.co/sBtDRxEnXJ

La newsletter DevSecOps
Joined November 2023
Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)Photo by Dele (@DeleLearnsTech)
๐—™๐—ผ๐—ฐ๐˜‚๐˜€ ๐—ผ๐—ป ๐—ช๐—ต๐—ฎ๐˜ ๐—ฌ๐—ผ๐˜‚ ๐—–๐—ฎ๐—ป ๐—–๐—ผ๐—ป๐˜๐—ฟ๐—ผ๐—น In the tech world, things change fast. Leaders often face unexpected challenges, be it a sudden change in client needs, new market trends, or internal team issues. It's easy to feel overwhelmed, but there's a simple idea that can help a lot: focus on what you can control. This can improve our happiness by an order of magnitude. Here are a few tips for focusing on what you can control: ๐Ÿ”น ๐—œ๐—ฑ๐—ฒ๐—ป๐˜๐—ถ๐—ณ๐˜† ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ฐ๐—ถ๐—ฟ๐—ฐ๐—น๐—ฒ ๐—ผ๐—ณ ๐—ฐ๐—ผ๐—ป๐˜๐—ฟ๐—ผ๐—น. What are the things that you have direct control over? This could include your thoughts, actions, reactions, attitude, and effort. It could also include your work environment, relationships, and finances. ๐Ÿ”น ๐—™๐—ผ๐—ฐ๐˜‚๐˜€ ๐—ผ๐—ป ๐˜†๐—ผ๐˜‚๐—ฟ ๐—ฎ๐—ฐ๐˜๐—ถ๐—ผ๐—ป๐˜€. Don't waste time and energy worrying about things outside your control. Instead, focus on taking positive steps to improve your situation. ๐Ÿ”น ๐—ฆ๐—ฒ๐˜ ๐—ฟ๐—ฒ๐—ฎ๐—น๐—ถ๐˜€๐˜๐—ถ๐—ฐ ๐—ด๐—ผ๐—ฎ๐—น๐˜€. Don't try to change everything at once. Set small, achievable goals that you can build on over time. ๐Ÿ”น ๐—–๐—ฒ๐—น๐—ฒ๐—ฏ๐—ฟ๐—ฎ๐˜๐—ฒ ๐˜†๐—ผ๐˜‚๐—ฟ ๐˜€๐˜‚๐—ฐ๐—ฐ๐—ฒ๐˜€๐˜€๐—ฒ๐˜€. When you achieve a goal, take the time to celebrate your accomplishment. This will help you stay motivated and focused on your plans. ๐Ÿ”น ๐—•๐—ฒ ๐—ธ๐—ถ๐—ป๐—ฑ ๐˜๐—ผ ๐˜†๐—ผ๐˜‚๐—ฟ๐˜€๐—ฒ๐—น๐—ณ. We often think wrong about ourselves and are too critical. But thinking and speaking friendly about ourselves can make a big difference in overall well-being. ๐Ÿ”น ๐—Ÿ๐—ฒ๐—ฎ๐—ฟ๐—ป ๐˜๐—ผ ๐˜€๐—ฎ๐˜† ๐—ก๐—ข (๐—ฎ ๐—น๐—ผ๐˜), especially for those activities that drain your energy. Choose wisely, and your productivity will skyrocket. Every "no" is a "yes" to something else you like more. As the new week starts tomorrow, remember to focus on things you can control. For things we cannot control, ๐˜„๐—ฒ ๐—ฐ๐—ฎ๐—ป ๐—ฑ๐—ฒ๐—ฐ๐—ถ๐—ฑ๐—ฒ ๐—ต๐—ผ๐˜„ ๐˜๐—ผ ๐—ฟ๐—ฒ๐—ฎ๐—ฐ๐˜ ๐˜๐—ผ ๐˜๐—ต๐—ฒ๐—บ. I wish you a great week ahead ๐Ÿ‘‹! Image: Jacqueline Rae Legard (Mindful Enough)
1
7
33
2,947
Reflected XSS: the payload sits in the URL and the server echoes it straight back into the page. The victim has to click a crafted link, and nothing is stored.
56
UPDATE: Web Security Academy is now built into Burp Suite. Your labs and progress will directly appear in the โ€˜Academyโ€™ tab in Burp whenever you open it. Launch any lab in Burp's browser, and its traffic flows into Proxy so you can use Repeater, Intruder, and the rest of the toolkit just as you would on a real target. Solved a lab? Burp updates your progress and suggests the next one on your learning path. - 273 labs across Apprentice, Practitioner, and Expert levels, grouped into learning paths - Burp AT's Academy mentor gives guiding questions based on your Proxy history without giving away the solution (For pro users only) - Free in Community Edition and Professional editions, with a free PortSwigger account Sign in once and your progress follows you. Available from November 17, 2026. portswigger.net/burp/pro/feaโ€ฆ
3
14
97
5,087
Stored XSS: the attacker saves a script once (a comment, a profile name) and your app serves it to every visitor. No link to click. Just open the page. Fix ideas: encode on output, sanitise rich text with an allowlist, add a CSP that blocks inline scripts.
1
57
Anthropic gave its new model old code with 1,829 real, known vulnerabilities still in it. It found 89.6% of them on its own. The test is called CVE Rediscovery. The model gets a snapshot of a project from before each bug was fixed and has to find the flaw. Anthropic's launch table: Mythos 5.2: 89.6% Opus 5.5 (safeguards off): 81.3% Mythos 5.1: 78.3% Three things to know before you share that number: 1. It's Anthropic's own internal test, so no other lab has a score. 2. These CVEs are public, so some fixes may already be in the training data. 3. Anthropic's own footnote says that on its other evaluations, bug-finding is "roughly on par with Opus 5.5". Finding a bug is step one. Exploiting it is a different test.
1
68
Anthropic's new model completes 76.3% of a simulated multi-stage cyberattack. Here's what that test actually involves. CyScenarioBench, from @Irregular, drops the AI into a fake company network. It has to pull passwords from a recorded meeting, trick simulated staff, hop between machines with stolen credentials and avoid tripping the alarms. Anthropic's launch table: Mythos 5.2: 76.3% Opus 5.5 (safeguards off): 67.6% Mythos 5.1: 61.7% GPT-6 Cyber: 61% Mythos is the same model as the public Fable 5.2, minus the safeguards that block most cyber tasks. The footnote matters: "each lab's scores are based on unique tasks." So, compare a lab with its own earlier models, not with other labs. teapawt.org/claudeai/status/211986โ€ฆ
Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges. Theyโ€™re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.
47
Namespaces decide what a container can see. Cgroups decide how much it can use. CPU, memory, disk I/O, process count. Docker's --memory and Kubernetes CPU and memory limits are cgroup settings underneath.
1
4
215
Meta's red team found an Apple bug that let iPhone and Mac apps record from the microphone even after the user denied permission. The app forged an ID token to pose as Apple's own audio service, which then skipped the permission check. On one iOS 27 beta, the orange mic dot never lit up, and the logs pinned the recording on Apple's service, not the app. Apple fixed it in August with iOS 26.6.1 and macOS Tahoe 26.6.2. Its patch note calls it just "a logic issue."
67
241
38
1,946
143,941
Clipping is here ๐ŸŽฌ Grab your favorite moment from any video, add your take, and post it without leaving X. Creators: every clip credits your original video and links back to it, so new viewers can find your work and keep watching. Clippers: you now have an easy, native way to share moments from your favorite creators directly on X. The best part? Both the original creator and the clipper can earn from views on the clip through Original Content Rewards. Everyone wins when great moments travel. Clipping is one more step toward making X the best place to create, grow and earn.
1,101
1,330
1,229
9,735
2,609,658
This command will save you more time than you think. At least it saved me ๐Ÿคฃ Kubectl config current-context
1
2
92
Learning AI engineering? These free resources are worth exploring: 1. huggingface.co/learn โ€” AI and LLM courses 2. kaggle.com/learn โ€” Python and ML exercises 3. fast.ai โ€” Practical deep learning 4. pytorch.org/tutorials โ€” PyTorch tutorials 5. scikit-learn.org โ€” ML documentation 6. docs.python.org โ€” Python reference You don't need to buy an expensive course before writing your first AI program.
16
170
4
1,005
25,882
Keep learning : You can ask your favorite agent to give you MCQ questions to ensure you understood what it built for you !
1
122
Dele retweeted
Changing one letter bypassed an entire admin API in @huggingface chat. One request could trigger bulk internal exports of every userโ€™s private chats, completely unauthenticated. Hereโ€™s how. ๐Ÿงต
30
129
55
1,458
146,530
Tip: Switch Claude Code's output style to Concise to get short, straight-to-the-point results. Enable it via /config โ†’ Output style, or add "outputStyle": "Concise" directly to your settings.json. You can still get the full details when you ask.
149
81
33
1,289
95,961
One thing Iโ€™ve learned working in DevOps is that understanding the infrastructure behind an application matters more than knowing how to use every new tool. Take time to understand how your applications communicate, how data flows, where failures can happen, and what happens when things go wrong. Learn to investigate incidents instead of just restarting services. Understand why infrastructure is designed a certain way, what it costs to maintain, and how to make it more reliable and secure. The more you understand the systems you manage, the better engineering decisions you make. Thatโ€™s real DevOps career growth.
3
12
67
1,666
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! ๐Ÿงต Blog: rikiyat.github.io/blog/weighโ€ฆ
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-spโ€ฆ github.com/RikiyaT/lm_undโ€ฆ huggingface.co/Qwen/Qwen3.5-0โ€ฆ huggingface.co/RikiyaT/Qwen3.โ€ฆ
2,368
7,253
5,757
55,903
10,099,030