Denoise · Twitter

The security conversation has moved from basic prompt injection to complex exploits in autonomous systems, with Anthropic and Google DeepMind both formalizing red-teaming practices.

Major labs like Anthropic and Google DeepMind are publicly releasing formal frameworks and disclosures for agent security, addressing jailbreaks and prompt injection.

• 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 battleground for coding assistants is moving from IDE extensions like Copilot to standalone terminal agents like Claude Code, with long-context task performance as a key differentiator.

Anthropic's release of Claude Code 1.5, a terminal-native coding agent, dominates the conversation, with prominent developers like @karpathy highlighting a fundamental workflow shift.

• 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 de facto agent stack is emerging, with OpenAI providing orchestration primitives, Anthropic's protocols being adopted by LangChain, and Vercel/Replit offering serverless deployment targets.

OpenAI, Vercel, and Replit all released tools for agent orchestration and deployment, signaling the rapid commoditization of the agent 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.

The focus remains on foundational data enablement, with Mistral AI continuing its pattern of releasing high-quality, open datasets to fuel community model development.

Mistral AI released a large-scale, 100M-row web OCR dataset, continuing its contribution of foundational resources for open model training.

• Mistral AI@MistralAIrising

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

The simple vector search of RAG is being challenged by more structured approaches like knowledge graphs (LlamaIndex) and dedicated memory layers (mem0ai) for agent state management.

Discussion focuses on the limitations of basic RAG and the challenges of massive context windows, promoting a shift toward more complex "context engineering."

• 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 principles of agentic automation are being integrated into established SaaS platforms like Notion and Linear, suggesting a broader trend of 'no-code' agentic workflows.

Workspace automation tools like Notion and Linear are shipping features that auto-triage and automate complex workflows, mirroring agent-like capabilities in SaaS products.

• 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 a craft into a science, with organizations like Weights & Biases running massive experiments to map the efficient frontier of prompt performance.

The focus is on systematic and large-scale benchmarking of system prompts to find optimal performance, moving beyond anecdotal tricks and tips.

• 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 agent training relies more on synthetic data, quality control and filtering techniques, as highlighted by @jerryjliu0, become critical for preventing model failures.

The key challenge discussed is dataset curation for agent training, specifically how to filter synthetic data to avoid poisoning model generalization.

• Jerry Liu@jerryjliu0repeated

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