yeah but it's hard to get that to be mor efficient than subagents.
PROMPT ENGINEERING
-
AI Model Excels at Text Rendering and Creative Generation
By
–
– I said that this was genuinely sad and I was trying to engage him around his oss efforts in a transparent way, undeserved reaction – As I said, it is not easy to be getting messages like from people I vouched for P.S. In the age of agents, the # of commits are irrelevant
-
Communicating with AI Agents: A Simple Approach
By
–
It's not that difficult, to be honest. Talking to your agents is all you need.
-
AI Agents Self-Improve by Rewriting Their Own Instructions
By
–
This is how AI agents self-improve in 2026. My OpenClaw AI agents rewrite their own instructions when something breaks and self-improve. Every Thursday, Kelly runs a self-review. She opens my profile, compares what she drafted vs what I actually posted, and measures the gap. Last week she found only 70% of her drafts shipped. She diagnosed the problem herself: category mismatch, not quality. Rewrote her own strategy file and killed an entire content format. I'm not in this loop. I didn't notice the problem. I didn't tell her what was wrong. I didn't write the fix. The agent compared her output against reality, diagnosed the gap, and rewrote her own instructions. That's the difference between an agent that follows orders and an agent that improve. Shubham Saboo (@Saboo_Shubham_) x.com/i/article/202179384677… — https://nitter.net/Saboo_Shubham_/status/2022014147450614038#m
→ View original post on X — @saboo_shubham_, 2026-03-29 19:19 UTC
-
Fine-tuning strategies and future plans for expanding capabilities
By
–
Thanks! The fine-tuning section is more general in the first one, and the second one is focused on math because it makes it easier from a security perspective. But I plan to add bonus materials over time, and function calling would be an interesting one.
-
AI Image Generation SVG Links and Content Wrapping Issues
By
–
But I mean it could be a link to a png, if you ask for a Mona Lisa svg you sometimes get a perfect Mona Lisa, which is just a link to Wikipedia inside a little code wrapper
-
RAG vs Fine-Tuning: What Works in Enterprise AI
By
–
RAG vs Fine-Tuning: What Actually Works in Enterprise AI (And Why Most Get It Wrong) https://
craigbrownphd.substack.com/p/rag-vs-fine-
tuning-what-actually?utm_source=dlvr.it&utm_medium=twitter
… #DataScience #DataAnalytics -
Hooks: Deterministic Control Over Tool Usage
By
–
hook are are another great way to control tool usage. They give you deterministic control over every tool call. I have covered them in the article.
-
Claude Skills: Self-Contained Workflow Packages for Efficient AI
By
–
What are Claude Skills?
— Akshay 🚀 (@akshay_pachaar) 29 mars 2026
𝗖𝗟𝗔𝗨𝗗𝗘.𝗺𝗱 was never meant to hold entire workflows.
But that's exactly where they end up. General rules, coding conventions, 20-step security review processes, deployment checklists. All in one file that loads into every single session, eating… https://t.co/fCLQezp4lG pic.twitter.com/VPLfFEYrTRWhat are Claude Skills? 𝗖𝗟𝗔𝗨𝗗𝗘.𝗺𝗱 was never meant to hold entire workflows. But that's exactly where they end up. General rules, coding conventions, 20-step security review processes, deployment checklists. All in one file that loads into every single session, eating context even when Claude is just renaming a variable. 𝗦𝗸𝗶𝗹𝗹𝘀 fix this by turning workflows into self-contained packages that Claude loads only when the task demands it. Here's the idea. A skill is a folder inside .𝗰𝗹𝗮𝘂𝗱𝗲/𝘀𝗸𝗶𝗹𝗹𝘀/. Each folder contains a 𝗦𝗞𝗜𝗟𝗟.𝗺𝗱 file with two things: a 𝗱𝗲𝘀𝗰𝗿𝗶𝗽𝘁𝗶𝗼𝗻 that tells Claude when to activate it, and the workflow instructions that tell Claude what to do. The description is the trigger. Claude reads all available skill descriptions, watches the conversation, and when your request matches, it pulls in that skill automatically. You don't paste the steps. You don't type a command. Claude recognizes the intent and invokes the right skill on its own. You can also trigger any skill explicitly with a slash command like /𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆-𝗿𝗲𝘃𝗶𝗲𝘄 when you want manual control. I recorded a deep dive on skills when they were first released, and everything in it is even more relevant today. The video below walks through exactly how this works. But auto-invocation is just the surface. The real power is what skills can carry with them. Skills are full packages, not just instruction files. A 𝗦𝗞𝗜𝗟𝗟.𝗺𝗱 can reference supporting files that live right next to it using the @ symbol. A detailed security standards document. A release notes template. A compliance checklist. Whatever the workflow needs, the skill bundles it together. Inside 𝗦𝗞𝗜𝗟𝗟.𝗺𝗱, YAML frontmatter defines the name, description, and which tools the skill is allowed to use. The 𝗮𝗹𝗹𝗼𝘄𝗲𝗱-𝘁𝗼𝗼𝗹𝘀 field is worth paying attention to. A security review skill only needs 𝗥𝗲𝗮𝗱, 𝗚𝗿𝗲𝗽, and 𝗚𝗹𝗼𝗯. It has no business writing files. Restricting tool access makes the skill safer and far more predictable. Skills live at two levels. Project skills go in .𝗰𝗹𝗮𝘂𝗱𝗲/𝘀𝗸𝗶𝗹𝗹𝘀/ and get committed to git so the whole team shares them. Personal skills go in ~/.𝗰𝗹𝗮𝘂𝗱𝗲/𝘀𝗸𝗶𝗹𝗹𝘀/ and follow you across every project. A 𝗖𝗟𝗔𝗨𝗗𝗘.𝗺𝗱 with a 20-step security process baked in is dead weight in 90% of your sessions. A 𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆-𝗿𝗲𝘃𝗶𝗲𝘄 skill that activates only when security is on the table is precision. 𝗖𝗟𝗔𝗨𝗗𝗘.𝗺𝗱 tells Claude what rules to follow. Skills tell Claude what workflows to execute. The article below is a complete guide to 𝗖𝗟𝗔𝗨𝗗𝗘.𝗺𝗱, hooks, skills, agents, and permissions, and how to set them up properly. Akshay 🚀 (@akshay_pachaar) x.com/i/article/203496196714… — https://nitter.net/akshay_pachaar/status/2035341800739877091#m
→ View original post on X — @akshay_pachaar, 2026-03-29 10:13 UTC