Denoise · Twitter

The focus of agent security is shifting from prompt injection to vulnerabilities in the orchestration layer, a concern echoed by @AnthropicAI, @GoogleDeepMind, and @AlexAlbert__.

Major AI labs are now publicly disclosing agent vulnerabilities and publishing red-teaming frameworks.

• 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 clear convergence is visible, moving from IDE copilots to autonomous terminal agents, a pattern established by @AnthropicAI's product and validated by @karpathy's analysis.

Anthropic's Claude Code 1.5 launch marks a significant shift toward terminal-native coding agents, with community benchmarks and analysis following.

• 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 stack is forming as @OpenAI, @LangChainAI, @vercel, and @replit all ship tools for orchestrating and deploying agents, solidifying this as a distinct infrastructure layer.

Infrastructure providers are racing to release SDKs, runtimes, and deployment harnesses for the new agent-native paradigm.

• 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 the conversation is on agents, @MistralAI's release shows a continued strategic investment in open, foundational multimodal datasets.

Mistral AI released a large, open dataset for web OCR, continuing its focus on providing foundational data for the community.

• Mistral AI@MistralAIrising

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

Practitioners like @GregKamradt and startups like @mem0ai are moving beyond basic vector search, signaling that agent complexity now demands more sophisticated memory architectures.

The discourse is evolving from simple RAG implementations to more complex 'context engineering' and structured memory systems 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.

While agent-native tools grab headlines, the integration of automation by @NotionHQ and @linear suggests a broader, quieter adoption of agent-like capabilities in existing SaaS platforms.

Established workspace tools like Notion and Linear are shipping AI-powered automation features for tasks like data entry and issue triage.

• 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 consolidating around data-driven methods, with large-scale experiments from @weights_biases representing a more rigorous approach than individual prompt engineering threads like @dotey's.

The focus is shifting from anecdotal prompt tricks to systematic, large-scale benchmarking of system prompts for optimal performance.

• 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 capabilities advance, the bottleneck is becoming data quality, with practitioners like @jerryjliu0 highlighting the challenge of identifying synthetic data that poisons generalization.

Attention is being paid to the nuances of dataset curation for agent training, specifically 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.