Denoise · Twitter

A convergence is visible between @AnthropicAI and @GoogleDeepMind on the need for formal agent security frameworks, shifting the focus from model-level jailbreaks to orchestration-layer exploits.

Major labs and security researchers are focusing on red-teaming autonomous agents, establishing frameworks for discovering and disclosing vulnerabilities.

• 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 direct competition is emerging between Anthropic's terminal agent model and the incumbent IDE-plugins like Copilot, with @levelsio and @swyx providing early evidence of a potential workflow disruption.

The launch of Anthropic's terminal-native Claude Code agent, supported by early benchmarks and user adoption, marks a significant shift in developer tooling.

• 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.

OpenAI, Anthropic, Vercel, and Replit are converging on a standardized architecture for agent deployment, solidifying a common pattern of local orchestration deploying to hosted, durable workers.

The ecosystem is rapidly building out a full-stack for agents, with major players releasing SDKs, orchestration tools, and specialized 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.

While a quiet day for model releases, @MistralAI's dataset contribution reveals the underlying infrastructure work required to advance multimodal capabilities across the industry.

Mistral AI released a large-scale, open web OCR dataset, providing a foundational resource for training 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 raw context windows (@reach_vb) and developing structured memory systems (@mem0ai, @GregKamradt), showing the path to effective agent memory is still contested.

The discussion on context is shifting from simple RAG implementations to more sophisticated 'context engineering' and dedicated memory layers 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.

The pattern here is the embedding of autonomous capabilities directly into SaaS products (@NotionHQ, @linear), abstracting away the complexity of agent orchestration for the end-user.

Workspace tools like Notion and Linear are shipping agent-like automation features, while Temporal provides a durable execution model for such tasks.

• 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 of @dotey (practical heuristics) and @weights_biases (large-scale empirical testing) show a convergence towards professionalizing prompt optimization as a data-driven discipline.

Prompt engineering is maturing from anecdotal tricks to systematic, large-scale benchmarking and optimization.

• 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.

The insight from @jerryjliu0 reveals that as agent capabilities scale, the bottleneck is shifting from raw data quantity to the sophisticated curation needed to ensure data quality.

A focus on data quality highlights the importance of filtering synthetic data to avoid poisoning agent generalization during training.

• Jerry Liu@jerryjliu0repeated

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