Denoise · Twitter

The critical attack surface is shifting from simple prompt injection to the agent's orchestration and tool-use layers, a focus for both @AnthropicAI and @GoogleDeepMind.

Frontier labs and security researchers are focusing on agent-specific vulnerabilities, releasing red-teaming frameworks and disclosures for complex attack chains.

• 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.

A direct battle for developer workflow is forming between @AnthropicAI's Claude Code and incumbents like Copilot, with community voices like @swyx and @levelsio already evaluating the practical trade-offs.

Anthropic's launch of Claude Code 1.5, a terminal-native agent, has spurred immediate benchmarks and debate around a fundamental workflow 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 clear convergence is happening towards a multi-worker, durable execution model for agents, with @OpenAI providing the SDK and platforms like @vercel and @replit providing the hosting environment.

OpenAI, Vercel, and Replit all released new infrastructure for deploying and orchestrating agents, signaling the emergence of standardized agent runtimes.

• 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 most of today's signals focus on agentic text systems, @MistralAI's data release provides foundational fuel for the next wave of multimodal agents that can process visual information.

Mistral AI contributed to the multimodal ecosystem by releasing a large, cleaned, and licensed web OCR dataset for training vision models.

• Mistral AI@MistralAIrising

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

Practitioners like @GregKamradt and projects like @mem0ai are fragmenting the concept of 'memory' into specialized components, moving beyond monolithic vector indexes to layered systems.

The discourse is evolving from 'RAG' to 'Context Engineering,' as developers share failure modes in massive context windows and explore new memory architectures.

• 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.

Agentic patterns are moving beyond developer tools into general productivity software, as both @NotionHQ and @linear abstract away complex user workflows into automated, multi-step actions.

Workspace automation tools are shipping agent-like features, with Notion and Linear launching capabilities for auto-filling data and auto-triaging issues.

• 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.

The community is shifting from anecdotal 'tricks' to data-driven optimization, with organizations like @weights_biases applying systematic benchmarking to find the efficient frontier for system prompts.

Prompt engineering is becoming more empirical, with practitioners sharing large-scale benchmark results and learnings from systematic reviews of production 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.

@jerryjliu0's comment highlights a critical, often-overlooked step in the MLOps lifecycle for agents: rigorous quality control for the synthetic data used in training.

A key challenge in training capable agents is the curation of synthetic data, specifically filtering out data that harms generalization.

• Jerry Liu@jerryjliu0repeated

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