Denoise · Twitter

The security conversation has shifted from the LLM endpoint to the orchestration layer, with @AnthropicAI and @GoogleDeepMind highlighting vulnerabilities in agent-tool interactions.

Major labs are focusing on red-teaming autonomous agents, releasing frameworks and disclosing complex jailbreaks beyond simple prompt injection.

• 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.

A consensus is forming around the terminal as the next developer interface, championed by @AnthropicAI and validated by influential developers like @karpathy and @swyx.

Anthropic's release of Claude Code 1.5, a terminal-native agent, is driving the conversation about a fundamental 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.

The industry is converging on durable, multi-worker architectures for agents, with OpenAI, Vercel, and Temporal all offering distinct but philosophically similar solutions for orchestration.

Major platforms, including OpenAI, Vercel, and Replit, are releasing SDKs and runtimes for deploying and orchestrating AI agents.

• 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.

Aside from a significant open dataset release from Mistral AI, this category was quiet, indicating a focus on agent infrastructure over new model modalities today.

Mistral AI contributed a large-scale, cleaned web OCR dataset to the open-source community for training multimodal models.

• Mistral AI@MistralAIrising

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

The discourse, led by figures like @GregKamradt, is shifting from retrieval to managing a memory hierarchy, a problem space now being addressed by tools like @mem0ai.

Developers are encountering the limits of simple RAG with large context windows, leading to new frameworks for "context engineering."

• 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 automation pattern seen in Notion and Linear parallels the agent trend, focusing on durable, chained workflows, similar to principles articulated by @temporalio for code.

Workspace automation tools like Notion and Linear are releasing features that automate routine tasks, mirroring the agent-driven trend in a SaaS context.

• 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.

Organizations like @weights_biases are industrializing prompt optimization, treating it as a large-scale search problem rather than a craft.

Prompt engineering is shifting from anecdotal tricks to systematic, large-scale benchmarking to find optimal system prompts.

• 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 point on filtering synthetic data highlights a key bottleneck: agent performance is increasingly limited by data quality, not just model size or architecture.

The discussion centered on the critical challenge of curating high-quality datasets for training robust and generalizable AI agents.

• Jerry Liu@jerryjliu0repeated

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