Denoise · Twitter

A consensus is forming that agent security is a critical, unsolved problem, with both @AnthropicAI and @GoogleDeepMind proactively publishing research to establish best practices.

The focus is squarely on red-teaming and securing autonomous AI agents, with major labs releasing frameworks and 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 conversation, led by @karpathy and validated by @AnthropicAI's launch, is converging on the terminal as the new primary interface for AI-native software development.

This category is dominated by the launch of Anthropic's Claude Code 1.5, a terminal-native agent changing 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.

Key players like @OpenAI, @vercel, and @replit are converging on a common set of primitives for agent orchestration and deployment, solidifying a new abstraction layer.

A wave of agent-focused infrastructure has been released, including SDKs, runtimes, and deployment harnesses from major platforms.

• 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 much of the industry focuses on agent orchestration, @MistralAI continues to invest in foundational data assets that enable the next generation of multimodal models.

The sole signal is a major open dataset release for web OCR from MistralAI, supporting multimodal model training.

• Mistral AI@MistralAIrising

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

Practitioners like @GregKamradt and @mem0ai are signaling that simple vector retrieval is insufficient for agents, pushing towards hybrid systems that include caches and knowledge graphs.

The discourse is shifting from basic RAG to more sophisticated 'context engineering' and structured memory for stateful 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 agentic patterns seen in AI-native tools are converging with SaaS workflows, as @NotionHQ and @linear automate complex, stateful operations previously done manually.

Workspace automation tools from Notion and Linear are adding agent-like capabilities for task management and triage.

• 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.

Prompt engineering is maturing into a rigorous discipline, with organizations like @weights_biases providing benchmark data that treats it as a hyperparameter tuning problem.

The focus is on moving from anecdotal prompt tricks to systematic, large-scale testing and optimization of 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 highlights a critical but often overlooked aspect of the agent stack: the data filtering and curation process required to prevent model poisoning and ensure generalization.

A focused discussion on the importance of curating high-quality synthetic data for training effective and safe agents.

• Jerry Liu@jerryjliu0repeated

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