Denoise · Twitter

The focus on agent security is proactive, with both Anthropic and Google DeepMind defining attack surfaces before widespread exploits emerge.

Major AI labs are publicly releasing red-teaming frameworks and responsible disclosures for agents, while practitioners are testing their real-world pentesting capabilities.

• 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 narrative, amplified by @karpathy and early adopters like @levelsio, positions terminal agents not as a supplement but a replacement for tools like VS Code + Copilot.

The launch of Anthropic's Claude Code 1.5, a terminal-native agent, is driving conversation about a fundamental shift in developer workflow away from traditional IDEs.

• 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 consensus architecture is emerging: a protocol for tool use (OpenAI SDK, MCP), a deployment harness (Replit), and a serverless runtime (Vercel Edge).

Platforms including OpenAI, Vercel, and Replit are shipping SDKs and runtimes specifically for deploying and orchestrating 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.

While the main conversation is on agents, major players like Mistral AI continue to invest in foundational data assets that power future model capabilities.

Mistral AI released a massive 100M-row web OCR dataset, providing a foundational, pre-cleaned resource for training next-generation multimodal models.

• Mistral AI@MistralAIrising

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

Frameworks from @GregKamradt and tools like @mem0ai suggest a split between volatile working memory and long-term storage, moving past reliance on vector search alone.

With 10M token context windows becoming a reality, the discussion is moving beyond simple RAG to 'context engineering' and specialized memory architectures 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 line between application-level automation (Notion, Linear) and general-purpose agent orchestration (Temporal) is blurring, showing a broad adoption of agentic patterns.

Workspace productivity tools like Notion and Linear are shipping agent-like automation features, while infrastructure providers like Temporal position their workflow engines for agent orchestration.

• 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 shift is from 'prompt engineering' as a craft to 'prompt optimization' as a systematic, data-driven process, exemplified by DSPy 3.0 and Weights & Biases' study.

The conversation around prompting is maturing from anecdotal tricks to systematic, large-scale benchmarking of system prompts, led by platforms like Weights & Biases.

• 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 agent training becomes more common, quality control for synthetic datasets, as discussed by @jerryjliu0, is becoming a critical and non-obvious challenge.

The focus is on the nuanced problem of data curation for agent training, specifically on techniques to filter out synthetic data that harms model generalization.

• Jerry Liu@jerryjliu0repeated

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