Denoise · Twitter

Major labs like Anthropic and Google DeepMind are formalizing agent security, shifting the attack surface from prompt injection to orchestration-level vulnerabilities.

The focus is on red-teaming autonomous agents, with new frameworks for prompt injection and analysis of real-world pentesting results.

• 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 consensus is forming around the terminal as the new IDE for AI-native development, a narrative accelerated by @karpathy and validated by early adopters like @levelsio.

Anthropic's Claude Code 1.5 launch dominates the conversation, positioning terminal-native agents as a direct challenger to the IDE/Copilot workflow.

• 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 platform race is consolidating between OpenAI, LangChain, and hosting providers like Vercel to define the protocol and runtime for multi-agent systems.

New SDKs and managed runtimes for deploying agents were released by OpenAI, Vercel, and Replit, signaling a push for standardized infrastructure.

• 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.

A quiet day for this category, with the only notable event being Mistral's foundational dataset release rather than new model or hardware capabilities.

Mistral AI released a large-scale, open web OCR dataset intended for training vision models.

• Mistral AI@MistralAIrising

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

Vector search is now seen as insufficient; developers like @GregKamradt and tools like @mem0ai and @llamaindex are exploring more complex memory and graph retrieval models.

Discussion shifts from basic RAG to 'context engineering,' focusing on cache invalidation at 10M token scale and more complex 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 seen in dedicated AI agent frameworks like Temporal are converging with features in mainstream SaaS productivity tools.

Workspace automation tools like Notion and Linear are shipping agent-like features for auto-filling tables and triaging issues.

• 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 into a data-driven optimization task, with firms like @weights_biases treating it as a hyperparameter search problem.

The focus is on moving from anecdotal prompt tricks to systematic, large-scale benchmarking of system prompts against frontier models.

• 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 quiet day, but @jerryjliu0's note reveals that dataset quality, not just quantity or architecture, remains a key bottleneck for advancing agent capabilities.

The challenge of curating high-quality synthetic data for agent training without poisoning model generalization was highlighted.

• Jerry Liu@jerryjliu0repeated

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