Denoise · Twitter

The focus of AI security is shifting from model evasion to agent orchestration vulnerabilities, with both @AnthropicAI and @GoogleDeepMind establishing best practices.

Major AI labs are now formally addressing the security risks of autonomous agents, publishing red team frameworks and responsible disclosures.

• 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 competition in AI coding is moving from IDE plugins (Copilot, Cursor) to standalone terminal agents, a shift articulated by @karpathy and benchmarked by @swyx.

Anthropic's release of Claude Code 1.5, a terminal-native agent, has triggered extensive discussion and benchmarks against existing tools.

• 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.

The stack is standardizing around agent orchestration. OpenAI's SDK is providing primitives while platforms like Vercel and Replit offer managed execution environments.

A wave of new infrastructure for deploying and managing agents has been released by OpenAI, Vercel, and Replit.

• 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.

This is a foundational data release aimed at improving a core multimodal capability, but the on-device and broader multimodal space was otherwise quiet.

Mistral AI released a large, clean, 100M-row web OCR dataset for training multimodal models.

• Mistral AI@MistralAIrising

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

Practitioners like @GregKamradt and tools like @mem0ai argue that a simple vector index is no longer sufficient, proposing more structured memory systems.

Discussion has moved from simple RAG to sophisticated 'context engineering' frameworks for managing agent memory and large context windows.

• 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 pattern of durable, background automation seen in developer tools like Temporal is now appearing in user-facing productivity apps like Notion and Linear.

Workspace automation features are becoming mainstream, with Notion and Linear launching agent-like capabilities for auto-filling and triaging.

• 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.

@weights_biases's large-scale study exemplifies a shift towards treating prompt engineering as a rigorous optimization problem rather than an art.

The focus is on systematizing prompt optimization, moving from anecdotal tricks to large-scale, data-driven benchmarking.

• 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 architectures mature, @jerryjliu0 highlights that the bottleneck is shifting from model design back to the classic ML problem of high-quality data pipelines.

The primary infrastructure concern for training agents is now dataset curation and filtering to avoid performance degradation.

• Jerry Liu@jerryjliu0repeated

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