Denoise · Twitter

The threat model is shifting from single prompt injections to complex, multi-step attacks on the agent's orchestration layer, a pattern visible in posts by @AnthropicAI and @GoogleDeepMind.

Frontier model labs are publicly disclosing agent-specific vulnerabilities and red-teaming frameworks as autonomous systems become more common.

• 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 consensus is forming around the terminal as the next major developer platform, championed by @AnthropicAI's Claude Code and validated by influential voices like @karpathy and @swyx.

The release of terminal-native coding agents is creating a clear divide between traditional IDE plugins and new, more autonomous developer workflows.

• 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 standard for deploying and managing agents is emerging, with @OpenAI, @vercel, and @replit all converging on a similar 'durable background worker' model for agents.

Infrastructure providers are racing to support new agent orchestration primitives from major labs, creating a new layer in the cloud-native stack.

• 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 is differentiating itself by contributing foundational, high-quality open data, a strategy that contrasts with the closed ecosystems of its primary competitors.

A major open dataset for web-based optical character recognition (OCR) has been released by a frontier model company.

• Mistral AI@MistralAIrising

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

A fragmentation of the RAG pattern is underway, with practitioners like @GregKamradt and tools like @mem0ai advocating for specialized layers for cache, working memory, and knowledge graphs.

The conversation is evolving from simple RAG implementations to more sophisticated 'context engineering' and complex memory architectures 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 trend shown by @NotionHQ and @linear indicates that complex agent orchestration is being abstracted away into simple, no-code automation features for end-users.

Workspace productivity tools are embedding agent-like automation features directly into their core workflows, such as auto-triaging issues and chaining database updates.

• 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.

A data-driven approach is consolidating, with platforms like @weights_biases enabling quantitative analysis that refutes common assumptions about what makes a good system prompt.

The practice of prompt engineering is maturing from sharing anecdotal tricks to conducting large-scale, systematic benchmarks 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.

@jerryjliu0's post reveals that for agent training, the key challenge is no longer data generation but rather sophisticated filtering to ensure data quality and prevent generalization failure.

Discussion in agent training is focused on the critical role of dataset curation and 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.