Denoise · Twitter

The conversation is shifting from theoretical prompt injection to securing complex, multi-tool autonomous agent systems, with red teaming becoming a standard practice.

Major labs Anthropic and Google DeepMind are releasing formal frameworks for agent security, while practitioners test autonomous agents for penetration testing.

• 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.

A split is emerging in developer tools between IDE plugins like Copilot and terminal-native agents like Claude Code, with the latter claiming higher developer velocity.

Anthropic's release of Claude Code 1.5, a terminal-native coding agent, is driving discussion, with early benchmarks and user testimonials appearing.

• 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 towards protocol-level standards is visible, with OpenAI's SDK and LangChain's integration work showing interoperability is a key concern.

OpenAI, Vercel, and Replit all shipped tools for agent orchestration and deployment, indicating a race to define the standard infrastructure layer.

• 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 orchestration is the focus, major labs like Mistral are still addressing foundational data bottlenecks for core multimodal skills like OCR.

Mistral AI released a large-scale, 100M-row web OCR dataset, providing a foundational resource for training multimodal models.

• Mistral AI@MistralAIrising

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

The field is advancing from basic retrieval to complex memory management, with tools like mem0.ai proposing new architectures that separate working and long-term memory.

The discussion highlights the failure modes of simple RAG in large context windows and argues for more sophisticated "context engineering" and 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.

SaaS tools like Notion and infrastructure like Temporal are converging on workflow orchestration, providing automation primitives at different layers of the stack.

Workspace automation is a prominent theme, with Notion and Linear releasing features to automate internal workflows, mirroring the broader agent trend.

• 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 maturing from anecdotal tricks to an industrial process, with platforms treating system prompt selection as a formal hyperparameter tuning problem.

The conversation covers both tactical prompt tricks from production reviews and strategic, large-scale system prompt benchmarking 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 capabilities scale, the bottleneck is shifting from model architecture to high-quality, poison-resistant training data, a classic ML pattern now repeating itself.

A key challenge highlighted is the curation of datasets for agent training, specifically how to filter 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.