Denoise · Twitter

The conversation is shifting from simple prompt injection to stateful, cross-tool exploits, with both Anthropic and GoogleDeepMind contributing to this new discipline.

Agent security is formalized with red-teaming frameworks and public disclosures, focusing on orchestration-level vulnerabilities.

• 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.

There's a clear convergence between Anthropic's product strategy, @karpathy's analysis, and user reports from @levelsio, consolidating the terminal as the next frontier for AI developer tools.

Anthropic's release of Claude Code 1.5 marks a major move towards terminal-native coding agents, a trend validated by key figures and early adopters.

• 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.

OpenAI's agent SDK and LangChain's MCP integration signal a push for protocol-level standards, while Vercel and Replit compete on the agent deployment and hosting layer.

The ecosystem is rapidly building the infrastructure for deploying and orchestrating agents, with major releases from OpenAI, Vercel, and Replit.

• 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 agent tools dominate the conversation, MistralAI's release demonstrates that foundational dataset curation remains a key competitive area for model builders.

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

• Mistral AI@MistralAIrising

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

The space is fragmenting into specializations: @GregKamradt advocates for context engineering over RAG, @mem0ai proposes structured memory layers, and @llamaindex focuses on graph retrieval.

The discussion around RAG is evolving into a more nuanced field of 'context engineering,' exploring memory architectures beyond simple vector retrieval.

• 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 ambient, automated task management in Notion and Linear suggests a broader trend of embedding agent-like capabilities directly into SaaS workflows.

Workspace automation and auto-triage features are being quietly shipped in major productivity tools like Notion and Linear.

• 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.

Both @weights_biases (via large-scale testing) and @dspy_ai (via compile-time optimization) are converging on the idea that optimal prompts should be discovered programmatically, not just written.

Prompt engineering is shifting from a manual art to a systematic science, with large-scale benchmarking and automated optimization frameworks gaining traction.

• 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 focus on filtering synthetic data reveals a key bottleneck in agent development: generalization is limited by data quality, not just model scale.

Conversation highlights the importance of careful dataset curation for training effective agents, particularly when dealing with synthetic data.

• Jerry Liu@jerryjliu0repeated

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