Denoise · Twitter

The focus of agent security is shifting from model-level prompt injection to vulnerabilities in the orchestration layer, a pattern highlighted by both @GoogleDeepMind and @AnthropicAI.

Major labs are formalizing autonomous agent security, releasing red-teaming frameworks and public disclosures on patched jailbreaks.

• 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 new product release, amplified by commentary from @karpathy, consolidates the narrative that developer workflows are undergoing a fundamental platform shift.

The release of Anthropic's Claude Code 1.5 has ignited discussion around terminal-native agents replacing 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 convergence is visible as both model providers (@OpenAI) and PaaS platforms (@vercel, @replit) are building competing primitives to become the standard for agent deployment.

The ecosystem is racing to provide the definitive infrastructure for agent orchestration, with new SDKs, protocols, and managed 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.

This release from @MistralAI continues their strategy of contributing foundational artifacts to the open-source community, targeting multimodal data processing.

Mistral AI released a large-scale, open dataset for web optical character recognition (OCR).

• Mistral AI@MistralAIrising

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

A fault-line is emerging between simple vector retrieval (RAG) and more complex memory architectures, with @GregKamradt and @mem0ai arguing for the latter.

The conversation is shifting from basic RAG to more complex 'context engineering' and structured memory systems 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 SaaS and AI is blurring as platforms like @NotionHQ and @linear embed automated triage and data-filling, competing directly with standalone AI wrappers.

Major workspace tools like Notion and Linear are quietly shipping powerful, agent-like automation features.

• 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.

A clear maturation of the practice is visible, moving from individual advice (@dotey) to industrial-scale analysis by platforms like @weights_biases.

The focus in prompt engineering is shifting from anecdotal tricks to systematic, large-scale benchmarking 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's point on synthetic data poisoning reveals a key bottleneck in scaling agent capabilities: data quality over data quantity.

The conversation highlights the critical need for careful curation of synthetic data for training robust agents.

• Jerry Liu@jerryjliu0repeated

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