Denoise · Twitter

A consensus is forming that the agent orchestration layer, not the core model, is the primary attack surface, a view shared by @GoogleDeepMind and @AnthropicAI.

Red teaming is now focused on autonomous agents, with major labs releasing frameworks and indie researchers testing real-world attack surfaces.

• 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.

@AnthropicAI's Claude Code is positioned as a direct challenger to the Cursor + Copilot paradigm, with early user reports from @levelsio and benchmarks from @swyx indicating a real shift.

The developer's primary interface is shifting from the graphical IDE to the terminal, driven by new conversational coding agents with file system access.

• 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 convergence pattern is visible across @OpenAI, @LangChainAI, @vercel, and @replit, who are all shipping components for a common agent deployment stack.

The infrastructure for deploying and managing agents is rapidly standardizing around orchestration primitives, protocol specifications, and dedicated edge runtimes.

• 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.

@MistralAI continues its strategy of commoditizing foundational resources to empower the open-source community against closed-source competitors.

A major open dataset for Optical Character Recognition (OCR) has been released to accelerate development of multimodal models.

• Mistral AI@MistralAIrising

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

A fault line is emerging between scaling context windows (@reach_vb) and building explicit memory layers (@mem0ai, @GregKamradt), suggesting RAG is no longer a one-size-fits-all solution.

Discussion is evolving from simple RAG towards 'context engineering,' focusing on sophisticated memory hierarchies, caching, and large context failure modes.

• 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 auto-triage work from @linear and chained updates from @NotionHQ show a convergence with the durable workflow concepts championed by platforms like @temporalio.

Workspace automation tools are shipping AI-driven features for auto-triaging and chained updates, mirroring patterns in durable execution frameworks.

• 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.

Work from @weights_biases and @dspy_ai refutes the 'prompt whisperer' narrative, consolidating around the idea that optimal prompts can be found computationally.

Production-grade prompt engineering is shifting from manual tricks and folklore 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 focus on synthetic data filtering reveals a key bottleneck in moving from pre-trained models to capable, specialized agents.

A critical but often overlooked aspect of agent training is the rigorous curation of synthetic datasets to prevent generalization failure.

• Jerry Liu@jerryjliu0repeated

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