Denoise · Twitter

The consensus from @AnthropicAI, @GoogleDeepMind, and @MalwareTechBlog is that agent security risks are moving up the stack from the model to the orchestration and tool-calling layers.

Red teams from Anthropic and Google are focused on securing autonomous agents, with early real-world tests showing vulnerabilities in the orchestration layer.

• 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 frontier is moving from IDE autocomplete (Copilot) to stateful, interactive terminal agents (Claude Code), while projects like DSPy focus on optimizing the underlying prompt logic.

Anthropic's Claude Code 1.5 launch establishes the terminal-native agent as a new developer tool category, with immediate benchmarks and adoption stories appearing.

• 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 pattern of convergence is visible: OpenAI is defining the protocols, while Vercel and Replit are providing the specialized serverless runtimes for hosting agent workers.

Major infrastructure providers like OpenAI, Vercel, and Replit are shipping SDKs and runtimes to support the deployment and orchestration of autonomous 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.

The release of a foundational dataset by MistralAI is significant, but the category is otherwise quiet, indicating the current market focus remains on text-based agents.

MistralAI released a large-scale, cleaned web OCR dataset to support the training of new 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 and projects like @mem0ai, now treats vector search as a solved commodity, focusing instead on higher-level memory architecture.

With context windows expanding to 10M tokens, discussion has shifted from simple retrieval (RAG) to sophisticated 'context engineering' and tiered memory systems.

• 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 shows AI evolving into an ambient orchestration layer within SaaS products, with platforms like Temporal providing the durable execution engines needed for such features.

Workspace tools like Notion and Linear are releasing AI-driven automation features that operate across their platforms, such as auto-triaging and chained database updates.

• 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.

Efforts by @weights_biases exemplify a move towards industrializing prompt optimization, treating it as a formal hyperparameter tuning problem rather than a craft.

The focus in prompt engineering is shifting from anecdotal tricks to systematic, large-scale benchmarking of system prompts to find an efficient frontier.

• 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.

As noted by @jerryjliu0, the bottleneck for improving agent performance is shifting from compute to sophisticated data filtering techniques that prevent generalization collapse.

The primary challenge discussed is the curation of high-quality training data for agents, specifically filtering out deceptive synthetic data.

• Jerry Liu@jerryjliu0repeated

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