Denoise · Twitter

The security focus is shifting from static LLM endpoints to complex, stateful agent systems, with Anthropic and Google DeepMind standardizing public discourse on agent jailbreaking.

Major AI labs, including Anthropic and Google DeepMind, are publicly releasing frameworks and disclosures on red-teaming autonomous agents.

• Anthropic@AnthropicAIrising

  Responsible disclosure on a Claude jailbreak chain we patched last week. Full write-up including our red team timeline.

• Google DeepMind@GoogleDeepMindrising

  New red team framework for prompt injection in autonomous agents. Covers cross-tool leakage, scanner evasion, and sandbox escape patterns.

• MalwareTech@MalwareTechBlogrepeated

  Autonomous agent running pentest flows against a real SaaS. First real-world run: fewer false positives than I expected on the vulnerability surface.

The developer tool landscape is rapidly moving towards terminal-based agents, with Anthropic's Claude Code positioning itself as a direct competitor to IDE-centric tools, a shift predicted by @karpathy.

Anthropic released Claude Code 1.5, a terminal-native agent, prompting immediate benchmarks and discussions about a paradigm shift in developer workflows.

• Anthropic@AnthropicAIrising

  Claude Code 1.5 is live. Terminal-native coding agent with full Claude Opus reasoning, file-ops sandbox, and session replay.

• Andrej Karpathy@karpathyrising

  The developer-experience shift from IDE to terminal agent is underrated. Coding workflows are about to look nothing like 2024.

• swyx@swyxrising

  Codex vs Claude Code terminal agent benchmarks. Pass@1 diverges more than I expected on the long-context editor tasks.

• DSPy@dspy_airising

  DSPy 3.0: prompt optimization via compile-time search over system prompt variations. Benchmarks inside.

• @levelsio@levelsiorising

  Switched my whole editor setup to Claude Code this week. Shipping faster than when I used Cursor + Copilot.

A new infrastructure stack for agents is converging, with OpenAI's agent SDK and protocols like MCP becoming central points of integration for platforms like Vercel and frameworks like LangChain.

OpenAI, Vercel, and Replit released new SDKs and runtimes for agent orchestration, while LangChain demonstrated integration with emerging agent protocols.

• OpenAI@OpenAIrising

  New agent SDK: protocol-level tool calling, deployment harness, and multi-worker orchestration primitives. Docs live.

• LangChain@LangChainAIrising

  MCP protocol integration thread. How to wire existing LangGraph agents into the Anthropic Model Context Protocol server spec.

• Vercel@vercelrising

  Edge runtime for agent workers is live. Spawn durable background agents from any serverless deployment.

• Alex Albert@AlexAlbert__rising

  When your security scanner finds nothing scary on an agent deploy, check the orchestration layer again. That's usually where the jailbreak sneaks through.

• Replit@replitrising

  New agent deployment harness. One command to go from local orchestration to hosted agent worker.

While agents dominate the conversation, foundational model providers like MistralAI continue to invest in open data as a key lever for community-driven model development.

MistralAI released a large, 100M-row web OCR dataset for open use, providing a new resource for training multimodal models.

• Mistral AI@MistralAIrising

  Open dataset release: 100M-row web OCR dataset. Cleaned, licensed, ready to train.

@GregKamradt is popularizing the term "context engineering" to supersede RAG, while startups like @mem0ai are building dedicated memory layers, signaling a new area of specialization in the agent stack.

The conversation is evolving from RAG to more sophisticated "context engineering" frameworks and multi-layered agent memory systems.

• Vaibhav Srivastav@reach_vbrising

  Tested the new 10M context memory window end to end. Surprising failure modes around rag retrieval cache invalidation, thread below.

• Greg Kamradt@GregKamradtrising

  RAG is dead, long live context engineering. My framework for when to cache, when to retrieve, and when to just dump memory into the prompt.

• mem0@mem0airising

  Memory layer for agents: differentiating working memory from the subconscious store. Vector index isn't enough anymore.

• LlamaIndex@llamaindexrepeated

  Knowledge graph retrieval walkthrough: when semantic vector search misses, graph hop beats it every time.

The concept of orchestration is a shared theme, appearing both in agent infrastructure via Temporal and in SaaS productivity tools like Notion and Linear, pointing to a broader trend.

Workspace automation features were launched by Notion and Linear, while Temporal positioned its durable workflow engine as a solution for orchestrating AI agents.

• Notion@NotionHQrising

  Notion workspace automation is out of beta. Auto-fill tables, chained updates across databases, and a new audit log surface.

• Linear@linearrising

  Linear now auto-triages incoming issues. Quiet launch, but already our favorite workspace feature of the year.

• Temporal@temporaliorepeated

  Orchestrating agents with durable workflows: replayable, resumable, and multi-worker by default. Walkthrough from our infra team.

• James Clear@jamesclearrepeated

  The best habit tracker is the one you actually open. Three open-source alternatives worth trying.

The field is moving from anecdotal prompt "tricks" to a data-driven science, with platforms like Weights & Biases providing the tooling to systematically optimize prompts at scale.

System prompt engineering is being treated more rigorously, with large-scale benchmarks from @weights_biases and practical guides from practitioners like @dotey.

• dotey@doteyrising

  Five prompt tricks learned this week from reviewing 200 production prompts. Short thread.

• Weights & Biases@weights_biasesrising

  System prompt benchmarking at scale: we ran 40k variants across 6 frontier models. The efficient frontier is not where you think.

As agent training scales, practitioners like @jerryjliu0 highlight that the key challenge is no longer data generation but filtering out synthetic data that poisons model generalization.

The focus in data preparation for agent training is shifting to the careful curation of synthetic data to avoid performance degradation.

• Jerry Liu@jerryjliu0repeated

  Dataset curation for agent training: how we filter synthetic data that looks good but poisons generalization.