Denoise · Twitter

Major labs like @AnthropicAI and @GoogleDeepMind are establishing agent security as a formal discipline, moving faster than the typical cat-and-mouse game.

Today's focus is on proactive security for autonomous agents, from red team frameworks to responsible disclosures of jailbreak chains.

• 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 consensus view, articulated by @karpathy and demonstrated by @AnthropicAI, is a structural shift away from IDE-based copilots toward conversational, stateful terminal agents.

Anthropic's Claude Code 1.5 release dominates, signaling a major push towards terminal-native coding agents that manage their own file state.

• 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 clear convergence is happening around agent orchestration, with @OpenAI providing official primitives while frameworks like @LangChainAI adapt to new protocols.

The ecosystem is building standardized agent deployment and orchestration layers, with major SDK and hosting releases from 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 single, high-quality dataset release from @MistralAI indicates a strategic focus on improving foundational vision-language capabilities across the open-source community.

MistralAI released a large-scale, 100M-row public dataset for web optical character recognition (OCR).

• Mistral AI@MistralAIrising

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

Practitioners like @GregKamradt and startups like @mem0ai are moving beyond basic vector search, suggesting RAG as a primitive is insufficient for stateful agents.

The conversation is shifting from simple RAG to more sophisticated 'context engineering' and multi-layered memory systems for agents.

• 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.

While not explicitly AI-branded, the automation features from @NotionHQ and @linear reflect a convergence on agent-like, autonomous behavior within SaaS workflows.

Workspace automation is a key theme, with Notion and Linear launching auto-triage and chained update features.

• 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 approaches from @weights_biases (large-scale benchmarks) and @dotey (manual review) show a maturing discipline of prompt engineering moving toward more rigorous, data-driven methods.

The focus is on systematic and large-scale benchmarking of system prompts to find optimal performance.

• 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.

@jerryjliu0's note reveals a critical, often-overlooked challenge in agent development: the subtle ways synthetic data can poison model generalization.

A specific insight was shared into the challenges of curating high-quality synthetic data for agent training.

• Jerry Liu@jerryjliu0repeated

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