Denoise · Twitter

Anthropic and Google DeepMind are converging on the need for formal, disclosed agent security frameworks, treating agent jailbreaking as a serious vulnerability class.

Major labs are publicly documenting their frameworks for red-teaming and securing autonomous agents, moving beyond theoretical risks to concrete practices.

• 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 conversation, led by voices like @karpathy and @swyx, positions terminal agents (Claude Code) in direct opposition to IDE-integrated assistants (Copilot), framing it as a paradigm shift.

Anthropic's release of Claude Code 1.5, a terminal-native agent, sparks discussion about a major shift in developer workflows 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.

There's a race to own the agent orchestration layer, with model providers like OpenAI and specialized infra platforms like Vercel and Replit offering competing primitives for deployment and management.

A wave of new tooling for agent orchestration and deployment arrives from OpenAI, Vercel, and Replit, solidifying the infrastructure layer for agent development.

• 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 competitors focus on proprietary agent frameworks, Mistral AI reinforces its strategy of providing open, foundational building blocks like large datasets for the community.

Mistral AI contributes to the open-source ecosystem by releasing a large-scale, cleaned web OCR dataset for training multimodal models.

• Mistral AI@MistralAIrising

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

A consensus is emerging, articulated by @GregKamradt and @mem0ai, that vector search alone is an inadequate memory model for agents, driving innovation in stateful memory architectures.

The discourse moves from simple RAG to sophisticated context engineering, exploring large-scale memory, cache invalidation, and structured memory layers 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 pattern shows a two-pronged approach to automation: new, standalone AI agents are being built, while existing productivity tools embed agent-like capabilities to augment established user habits.

Established SaaS platforms like Notion and Linear are rolling out autonomous features that automate complex internal workflows like issue triage and 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.

The field is maturing on two fronts: practical, hands-on advice from engineers like @dotey coexists with industrial-scale, data-driven analysis from platforms like @weights_biases.

Efforts to systematize prompt engineering continue, with practitioners sharing heuristics while research-oriented teams publish large-scale benchmark results.

• 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 articulated by @jerryjliu0, the critical bottleneck for advancing agent capabilities is no longer just model size or architecture, but the subtle art of dataset curation to prevent generalization poisoning.

The focus in agent training infrastructure is shifting toward the complex challenge of curating high-quality 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.