Denoise · Twitter

The security focus is shifting from simple prompt injection to complex agent orchestration vulnerabilities, as seen in disclosures from Anthropic and Google DeepMind.

Major AI labs are now publicly disclosing red team frameworks and patched jailbreaks for autonomous agents.

• 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 battle for AI coding assistants is moving from IDE plugins (Copilot) to standalone terminal agents (Claude Code), with early benchmarks from @swyx showing performance divergence.

Anthropic launched Claude Code 1.5, a terminal-native agent, prompting widespread discussion about the shift away from traditional IDEs.

• 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 new infrastructure layer is solidifying around agent orchestration, with OpenAI and LangChain converging on standardized protocols for multi-agent systems.

Major infrastructure providers like OpenAI, Vercel, and Replit are releasing SDKs and runtimes for deploying and orchestrating 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.

Mistral AI's single, large-scale data release highlights that access to high-quality, openly licensed data remains a key driver for advancing foundational multimodal capabilities.

Mistral AI contributed a large-scale, open web OCR dataset, providing a significant new resource for training multimodal models.

• Mistral AI@MistralAIrising

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

Practitioners like @GregKamradt and tools like @mem0ai are moving beyond vector search, proposing frameworks for managing complex memory as context windows grow.

The conversation is shifting from simple RAG to 'context engineering,' focusing on memory hierarchies and advanced retrieval strategies.

• 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 trend of embedding agent-like automation directly into SaaS tools continues, with Notion and Linear following a pattern of making user workflows self-managing.

Workspace automation is a key theme, with Notion and Linear launching features for auto-filling and auto-triaging business data.

• 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 a data-driven discipline, with platforms like Weights & Biases enabling large-scale experiments to quantify prompt performance.

Discussions center on systematic and large-scale benchmarking of system prompts to find optimal configurations.

• 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 model capabilities advance into agentic tasks, data quality becomes the bottleneck, with practitioners like @jerryjliu0 developing new filtering techniques for synthetic datasets.

The focus is on the specialized data curation required for training agents, specifically on filtering harmful synthetic data.

• Jerry Liu@jerryjliu0repeated

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