Denoise · Twitter

The threat model is shifting from prompt injection to orchestration-layer exploits, a convergence point for research from @GoogleDeepMind and disclosures from @AnthropicAI.

Major AI labs are proactively releasing red-teaming frameworks and vulnerability disclosures specifically 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 'terminal as the new IDE' pattern is consolidating, with @karpathy providing the thesis and @AnthropicAI delivering a product that directly competes with Copilot.

Anthropic's release of Claude Code 1.5, a terminal-native coding agent, dominates the conversation, sparking benchmarks and workflow discussions.

• 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 consensus architecture for agent deployment is emerging around durable background workers, supported by @OpenAI's SDK and @vercel's Edge runtime.

A wave of new tools from OpenAI, Vercel, and Replit provide the primitives for deploying and orchestrating 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.

While others focus on agent tooling, @MistralAI continues to provide foundational open-source data artifacts, reinforcing its strategy of enabling the broader research community.

MistralAI released a large, clean, 100M-row web OCR dataset for training multimodal models.

• Mistral AI@MistralAIrising

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

With massive context windows now available, the focus shifts to managing that context effectively, with players like @mem0ai and @llamaindex addressing the scaling limits of simple vector search.

The discussion is evolving from simple RAG towards 'context engineering,' exploring memory hierarchies, caching, and graph 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 agentic automation pattern is entering mainstream SaaS, with @NotionHQ and @linear implementing features that mirror the autonomous task completion seen in developer-focused agents.

Productivity SaaS tools like Notion and Linear are shipping AI-powered automation features for workspace management and issue triage.

• 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 discipline, with platforms like @weights_biases enabling rigorous optimization that moves beyond the trial-and-error approach shared by creators like @dotey.

The focus is shifting from anecdotal prompt tricks to systematic, large-scale benchmarking of system prompts to optimize model performance.

• 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 point reveals that as agent models become more capable, the bottleneck returns to data quality, specifically avoiding generalization poisoning from low-quality synthetic data.

Conversation centers on the challenges of dataset curation for agent training, particularly filtering synthetic data to avoid performance degradation.

• Jerry Liu@jerryjliu0repeated

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