Denoise · Twitter

The focus in agent security is shifting from prompt-level attacks to vulnerabilities in the orchestration layer, a pattern noted by @AnthropicAI, @GoogleDeepMind, and @AlexAlbert__.

Major labs are publicly disclosing agent jailbreaks and releasing formal red-teaming frameworks for autonomous systems.

• 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 ecosystem is moving to quantify the performance of new terminal agents like Claude Code against incumbents, with @swyx providing benchmarks and @levelsio offering anecdotal adoption reports.

Anthropic's release of Claude Code 1.5, a terminal-native agent, has triggered extensive discussion and benchmarking against existing AI coding tools.

• 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 convergence is happening around the agent deployment stack, with @OpenAI providing SDKs, @vercel offering runtimes, and @LangChainAI working on protocol interoperability.

OpenAI, Vercel, and Replit all launched new infrastructure for deploying and orchestrating agents, from SDKs to 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.

Although a quiet day for the category, @MistralAI's open dataset release is a significant infrastructural contribution for training future open-source multimodal models.

Mistral AI released a large, 100M-row web OCR dataset for public use in model training.

• Mistral AI@MistralAIrising

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

Practitioners like @GregKamradt are reframing the problem beyond RAG, while platforms like @mem0ai and @llamaindex are building solutions for more complex memory and knowledge graph retrieval.

The conversation shifts from basic RAG towards 'context engineering,' focusing on cache invalidation in large contexts and advanced memory architectures.

• 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 orchestration patterns for software agents, exemplified by @temporalio, are now being productized for knowledge workers inside tools like @NotionHQ and @linear.

Workspace automation tools from Notion and Linear are launching, echoing the durable workflow patterns discussed by infrastructure providers like Temporal.

• 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 craft of prompt engineering is maturing from anecdotal tricks to data-driven science, demonstrated by @weights_biases's large-scale benchmark study.

Focus is on systematic, large-scale benchmarking of system prompts and sharing concrete, production-tested prompt optimization techniques.

• 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.

A single post from @jerryjliu0 highlights a deep problem in the MLOps for agents stack: filtering out synthetic data that harms generalization, a non-obvious but crucial step.

A key discussion point is the subtle but critical challenge of curating high-quality synthetic data for training effective agents.

• Jerry Liu@jerryjliu0repeated

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