Denoise · Twitter

A clear pattern is emerging where major labs like Anthropic and Google DeepMind are publicly building security practices in parallel with agent capability development.

The security community is proactively developing frameworks and disclosing vulnerabilities for the new class of autonomous AI 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.

Competition in the agent coding space is consolidating around terminal-native workflows, with @karpathy validating the UX shift and @swyx providing early benchmarks.

The launch of Anthropic's Claude Code 1.5 signals a major push toward making the terminal the primary interface for AI-assisted development.

• 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.

There's a rapid convergence from OpenAI, Vercel, and Replit on providing dedicated infrastructure for hosting and managing background agent processes.

Major infrastructure players are now shipping primitives for deploying and orchestrating agents, solidifying the 'agent worker' as a standard compute unit.

• 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.

MistralAI continues to pursue a strategy of releasing high-quality, open data artifacts, differentiating itself from the API-first approach of its main competitors.

The category is quiet today, with the only major signal being a large-scale open dataset release for OCR from Mistral AI.

• Mistral AI@MistralAIrising

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

A new vocabulary is solidifying around advanced context management, with figures like @GregKamradt leading the conceptual shift while tools like LlamaIndex and mem0ai build the primitives.

The conversation is evolving from simple RAG to more sophisticated 'context engineering', focusing on caching, retrieval strategies, and multi-layered memory.

• 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.

A convergence is visible between dedicated agent orchestrators like Temporal and SaaS platforms like Notion/Linear, which are both building durable, automated workflow systems.

Workspace automation tools from Notion and Linear are adding agent-like capabilities, suggesting a broader trend of embedding autonomous features into SaaS.

• 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.

Platforms like Weights & Biases are enabling a more empirical approach to prompt optimization, validating the best practices shared by practitioners like @dotey.

The focus is on moving prompt engineering from an art to a science through 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.

The insight from @jerryjliu0 suggests that as model architectures advance, the frontier of agent performance is increasingly defined by data quality and curation techniques.

A single signal highlights that dataset curation, especially filtering synthetic data, is a critical bottleneck for training capable agents.

• Jerry Liu@jerryjliu0repeated

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