Denoise · Twitter

Agent security is moving beyond simple prompt injection to complex vulnerabilities in orchestration and tool-use, a focus shared by Anthropic's disclosure and Google DeepMind's framework.

Major model providers are publicly disclosing agent jailbreaks and releasing red-teaming frameworks, while independent researchers test autonomous pentesting.

• 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 frontier of AI coding assistants is shifting from IDE extensions like Copilot to standalone terminal agents like Claude Code, a paradigm change articulated by @karpathy.

Anthropic's release of Claude Code 1.5, a terminal-native agent, dominates the conversation, spurring immediate benchmarks and declarations of workflow shifts.

• 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 stack for agents is standardizing, with OpenAI's SDK and LangChain's protocol integrations suggesting a move towards interoperable orchestration.

OpenAI, Vercel, and Replit all released primitives for agent orchestration and deployment, indicating a rapid build-out of this new 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.

A quiet day, with Mistral's dataset release being the only signal, reinforcing that progress in multimodality remains fundamentally gated by large, clean datasets.

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

• Mistral AI@MistralAIrising

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

As context windows expand to 10M tokens, the problem shifts from retrieval to management, with practitioners like @GregKamradt and tools like LlamaIndex exploring more complex memory and graph-based systems.

Discussion is shifting from simple RAG towards 'context engineering'—managing large memory stores and complex retrieval strategies for agents.

• 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 seen in developer tools is being mirrored in business applications by Notion and Linear, automating triage and database updates.

Productivity tools like Notion and Linear are shipping agent-like automation features, suggesting these patterns are moving into mainstream 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.

Prompting is shifting from a craft of individual tricks (@dotey) to a science of large-scale benchmarking, exemplified by Weights & Biases' 40k variant study.

The focus is on systematizing prompt optimization by running large-scale benchmarks to find the most effective system prompt variations.

• 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.

For advanced agent training, data quality is now a more significant bottleneck than data quantity, requiring sophisticated filtering of synthetic data as noted by @jerryjliu0.

The key challenge discussed is the curation of high-quality synthetic data for agent training, focusing on avoiding data that harms generalization.

• Jerry Liu@jerryjliu0repeated

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