The MariaDB Foundation is hosting its first meetup for the Indian MariaDB community. It is online, so you can join from anywhere in the country.
An evening of talks and conversation with MariaDB…
Please RSVP on the Monkeys platform: https://t.co/R1yPvLmyoP
@mariadb
Google just open-sourced Harness².
(turn any harness into a self-improving system)
a harness is everything wrapped around the model, like the system prompt, tools, memory, skills, and subagents. today a person designs it once, and the agent runs every new task with that same fixed setup.
Harness² lets the agent refine that setup for each new task on its own, with the model weights left untouched.
on every task, the agent proposes edits to its harness, runs a few candidate versions, and composes a better harness by comparing how those runs behaved.
because it learns from the differences between runs, it needs no correct answer or grader for the task.
that covers a single task. Harness² also carries what it learns from one task into the next.
after each task, it saves the harness it settled on along with the lessons that held up across the runs. the next time it refines a harness, it starts from those notes instead of from scratch.
this means every finished task makes the next round of edits more informed.
Google tested this with Gemini 3.5 Flash and GLM-5 on legal, office, and professional work benchmarks. one round of edits was enough to beat their original harnesses, and giving the agent more rounds pushed scores higher.
those extra rounds can go wide or deep. the agent either tries several candidate harnesses at once or keeps refining a single one step by step, using OpenCode or Codex to carry out the work.
you can find the Harness² repo here: https://t.co/i00WLeeFpw
I wrote about a similar idea for agents already running in production, where the harness should repair itself after a failure instead of waiting for someone to debug the trace by hand.
that piece walks through how @Cometml's Opik (open-source) closes this loop, going from a failing trace to a proposed fix to a regression test that stops the same failure from coming back. it's quoted below.
THIS IS ACTUAL F*CKING GOLD
Anthropic just quietly dropped one of the best setups for Claude Code
The rule: orchestrate Sonnet, Haiku and Opus 5.5 simultaneously, locking every model into the exact lane it dominates
Opus-grade intelligence on critical architectural calls, zero Opus burn on routine terminal iterations
Launch Claude Code with this flag pair:
claude --model sonnet --advisor opus
how the workload divides:
→ Sonnet 5.5 leads the main session and cuts the actual code
→ Haiku 5.5 swarms repos, greps call sites, and parses docs concurrently
→ Opus 5.5 remains on standby as advisor for high-leverage checkpoints
→ you eliminate token waste on work that never required frontier reasoning
Opus only gets triggered on three specific events:
→ right before locking an architectural plan
→ the second an identical error hits twice
→ right before marking a complex task complete
Sonnet builds · Haiku explores · Opus reviews
JEV engineering applies the exact same logic one layer deeper: deterministic branches (file selection, tool dispatch, fast retry vs abort) clear in sub-milliseconds, so the frontier models only handle genuine forks
the architecture tree:
> Sonnet 5.5 drives the primary terminal session and edits code
> search crawls repo structure on Haiku 5.5
> grep locates call signatures on Haiku 5.5
> docs fetches API specs and schemas on Haiku 5.5
> Opus 5.5 stands on call as advisor, ingesting full context when pinged
> JEV resolves instant deterministic routing
feed this tree and prompt directly into Claude Code ↓
"Configure my Claude Code environment around this multi-model tree:
1. Run the primary session on Sonnet 5.5 with Opus 5.5 as advisor
2. Spawn three read-only Haiku 5.5 subagents: search (repo mapping), grep (call sites), docs (API references)
3. Enforce one rule in CLAUDE.md: ping advisor before locking any plan, upon repeating any error twice, and before final task handoff
4. Route instantaneous deterministic forks (tool choice, path check, retries) through JEV
Show every file update as a diff first, make zero changes without confirmation"
↳ https://t.co/gvAWk6JDNP
bookmark this, you'll run it every single day
Thank you for making my Network Basics for Hackers #1 bestseller among Linux Applications.
This timeless classic continues to be a favorite among those new to cybersecurity and networks.
I am very grateful 🙏
https://t.co/qowc0GwxGI
Community Features Are Here!
Finding your people and growing your network just got a whole lot easier. We’re thrilled to introduce our brand-new Community Features, designed to bring you together with like-minded people online and in person!
Here’s a quick look.
After mentioning this Context Engineering guide on the podcast, many of you asked for it, so I’m happy to finally share it :)
https://t.co/sh2RO9mTBi
This 21-page guide is enough to help you understand how memory, retrieval, agents, chunking, query rewriting, and tools actually work together.
I put it together as a practical look at how to give AI systems the right information, at the right time, and build systems that work reliably in the real world.
Hope you find it useful. I’d love to hear what you build with it.
SSH Tunnels or How to Become a Connectivity Wizard 🧙♂️
10 hands-on challenges to practice port forwarding with SSH tunnels - starting with simple local/remote port forwarding and all the way down to advanced scenarios with bastions and jump hosts:
- Access an internal debug port through an SSH tunnel https://t.co/jlaEw8y7zy
- Reach a private VPC service through an SSH bastion https://t.co/FCNs9iFojr
- Reach a remote loopback port through an SSH jump host https://t.co/peLYwVG0SJ
- Restrict SSH bastion access by user role https://t.co/74s3isK2pY
- Reach internal servers through a "no shell" SSH bastion https://t.co/3dSvgKPoJJ
- Reach a whole VPC through an SSH SOCKS proxy https://t.co/CVbU9lqQKP
- Expose a local service through an SSH reverse tunnel https://t.co/fgXYIVUWVT
- Expose a home network device through an SSH reverse tunnel https://t.co/TSSLIHIebg
- Expose a whole home network through an SSH reverse SOCKS proxy https://t.co/9KfDcqR1eM
- Replace password-based root access with a key-based admin login https://t.co/l3vK7dX5ao
Happy hacking!
RAG is not just one technique, it is an entire ecosystem of intelligence.
From context-aware assistants to domain-specific systems, here are 16 types of RAG models shaping the next wave of AI innovation -
1. Standard RAG
The foundation of all RAG systems - combines retrieval and generation for question answering and knowledge synthesis.
2. Agentic RAG
Empowers AI agents to retrieve and act autonomously, perfect for assistants that need dynamic, tool-based reasoning.
3. Graph RAG
Uses knowledge graphs for relational reasoning - ideal for expert systems in law, medicine, and semantic search.
4. Modular RAG
Breaks retrieval, reasoning, and generation into independent components - enabling collaborative, scalable AI workflows.
5. Memory-Augmented RAG
Adds persistent external memory for context retention, powering long-term chatbots and personalized experiences.
6. Multi-Modal RAG
Processes text, images, and audio together - perfect for video summarization, captioning, and multi-modal AI tools.
7. Federated RAG
Enables privacy-preserving retrieval from decentralized sources, used in healthcare and secure enterprise systems.
8. Streaming RAG
Performs real-time retrieval and generation, ideal for financial dashboards, live feeds, and social media monitoring.
9. ODQA RAG (Open-Domain QA)
Handles large, diverse datasets - ideal for search engines and intelligent virtual assistants.
10. Contextual Retrieval RAG
Maintains session-level awareness, great for conversational AI and customer support chatbots.
11. Knowledge-Enhanced RAG
Integrates structured domain data, useful for legal, educational, and professional knowledge applications.
12. Domain-Specific RAG
Custom-tailored for specific industries - like finance, healthcare, or legal analytics.
13. Hybrid RAG
Combines multiple retrieval approaches, bridging structured and unstructured data for high precision.
14. Self-RAG
Introduces self-reflection to refine its own answers, enabling AI models to fact-check and improve reasoning autonomously.
15. HyDE RAG (Hypothetical Document Embeddings)
Generates hypothetical documents to guide retrieval, excellent for complex or niche query contexts.
16. Recursive / Multi-Step RAG
Performs multiple retrieval-generation loops, enabling advanced problem-solving and reasoning chains.
From simple retrievals to self-improving AI reasoning loops, RAG is evolving fast.
Which type do you think will dominate enterprise AI systems in 2026?
Graph Convolutional Network by hand ✍️ ~ 12 steps walkthrough below
Graph Convolutional Networks (GCNs), introduced by Thomas Kipf and Max Welling in 2017, are the tool for data shaped like a graph: social networks, recommendations, biological networks, drug discovery, molecular chemistry.
I drew and calculated a simple GCN entirely by hand.
Goal: run a two-layer GCN, then a small classifier, on a five-node graph, filling in every cell yourself.
1. Given
A graph of five nodes, A to E, with edges between some of them.
2. Adjacency matrix (neighbors)
Put a 1 wherever two nodes share an edge, in both directions.
3. Adjacency matrix (self)
Add 1s down the diagonal, one self-loop per node. That is just adding the identity matrix.
4. Messages
Multiply each node's embedding by the weights and biases, then ReLU. Negatives become 0.
5. Pooling
Multiply the messages by the adjacency matrix. Each node gathers the messages of its neighbours and itself.
6. Visualize
Node A pools [3,0,1] + [1,0,0] = [4,0,1].
7. Second GCN layer
Messages again: weights, biases, ReLU.
8. Pooling again
Pool over each node and its neighbours, once more.
9. Visualize
Node C pools [1,2,4] + [1,3,5] + [0,0,1] = [2,5,10].
10. Fully connected layer
Weights, biases, ReLU. This time there are no neighbours to pool, just the node itself.
11. Linear layer
One more: weights and biases.
12. Sigmoid
Squash each score to a probability (≥ 3 → 1, 0 → 0.5, ≤ -3 → 0). That is the classification for each node.
You have just classified every node in the graph by hand. ✍️
The outputs:
A: 0 (very unlikely)
B: 1 (very likely)
C: 1 (very likely)
D: 1 (very likely)
E: 0.5 (neutral)
The takeaway: a GCN layer is two parts. The top part pools each node with its neighbours through the adjacency matrix. The bottom part is an MLP that transforms each node on its own. A transformer layer has the same two parts, with an attention matrix where the adjacency matrix was. Both matrices do one job, mixing across positions: attention over tokens, adjacency over nodes.
In my class I call the GCN the transformer's little cousin: a bit more stubborn, because its attention is fixed by the graph rather than computed from Q, K, and V. Draw the two side by side and the resemblance is hard to miss.
💾 Save this post!
#AIbyHand #GraphNeuralNetworks #DeepLearning
You don't need a CS degree to become an AI engineer
You need 6 months of building and one roadmap that actually tells you what to build
This is the roadmap
this is pure f*cking treasure
how to build your first ai agent (complete walkthrough)
if someone handed me this a year ago, i would’ve launched my first agent in a single day instead of burning two weeks
with the right person behind it, this rewrites the whole game:
try: https://t.co/4rEItVdKAW