Denoise · Twitter

The frontier of agent security is shifting from prompt injection to orchestration-level vulnerabilities, with major labs like @AnthropicAI and @GoogleDeepMind now publishing formal frameworks.

Discussions focus on advanced red-teaming techniques for autonomous agents, moving beyond basic prompt security.

• 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 primary interface for coding assistants is fragmenting, with @AnthropicAI's terminal-native agent presenting a direct challenge to the IDE-centric model, a shift validated by @karpathy.

Anthropic's launch of Claude Code 1.5 dominates, framing the day's conversation around a shift to terminal-native coding agents.

• 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.

The agent infrastructure stack is converging around orchestration primitives, with @OpenAI, @vercel, and @replit all shipping tools that point towards a common, hosted agent worker model.

Major players are releasing SDKs and runtimes aimed at standardizing the deployment and orchestration of AI 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 agent narrative is dominant, @MistralAI continues to execute a differentiated strategy by providing foundational, open datasets, strengthening the data layer of the ecosystem.

MistralAI released a large-scale, open web OCR dataset for training multimodal models.

• Mistral AI@MistralAIrising

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

A consensus is forming that simple vector retrieval is insufficient; actors like @GregKamradt and @mem0ai are fragmenting the RAG pattern into more sophisticated memory architectures.

The conversation is evolving from simple RAG to more complex 'context engineering' frameworks and layered 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 pattern of agentic automation is converging across both developer tools and general productivity software, with @NotionHQ and @linear's releases reflecting the same underlying trend.

Major productivity tools like Notion and Linear are shipping agent-like workspace automation features, moving them out of beta.

• 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 industrializing. The work by @weights_biases and the approach of @dspy_ai show a convergence towards treating prompt optimization as a data-driven, compiler-like problem.

The focus is on moving from anecdotal prompt 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.

As agent capabilities grow, @jerryjliu0's comment reveals that the bottleneck is shifting to data quality control, specifically filtering out plausible but harmful synthetic data.

A single signal highlights the difficulty of curating synthetic data for agent training without poisoning model generalization.

• Jerry Liu@jerryjliu0repeated

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