Denoise · Twitter

A consensus is forming among @AnthropicAI, @GoogleDeepMind, and practitioners like @MalwareTechBlog that agent security vulnerabilities lie in the orchestration layer, not just the model.

The focus is on establishing formal frameworks and real-world tests for red-teaming autonomous agents, moving beyond simple prompt attacks.

• 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 terminal is solidifying as the next major developer interface, with @AnthropicAI's launch serving as a catalyst for adoption (@levelsio) and technical analysis (@swyx).

Anthropic's release of Claude Code 1.5 dominates the conversation, driving discussions on terminal-native workflows and performance benchmarks against competitors.

• 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 platform race is visible between @OpenAI, @vercel, and @replit to provide the definitive agent hosting layer, while protocols mentioned by @LangChainAI suggest a push for interoperability.

Major infrastructure providers are shipping primitives for deploying, orchestrating, and hosting agents, signaling a race to own the agent deployment 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.

While most of the attention is on agents, @MistralAI's dataset release reinforces their distinct strategy of commoditizing model components through open-source contributions.

Mistral AI contributed a large, open dataset for web-based optical character recognition, continuing its strategy of releasing foundational resources.

• Mistral AI@MistralAIrising

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

A fault line is emerging between scaling context windows (@reach_vb) and building explicit, structured memory layers (@mem0ai, @GregKamradt), suggesting RAG alone is insufficient for complex agents.

The conversation is evolving from simple RAG to more sophisticated 'context engineering' frameworks that differentiate between types of agent memory.

• 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 autonomous workflow patterns seen in AI-native tools are now appearing in SaaS products from @NotionHQ and @linear, using principles similar to durable execution platforms like @temporalio.

Mainstream productivity tools are launching agent-like automation features, while durable execution frameworks provide the underlying primitives.

• 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 into a data-driven optimization problem, with platforms like @weights_biases providing tooling for the systematic search that goes beyond individual tips from creators like @dotey.

The focus is on moving from anecdotal prompt tricks to systematic, large-scale benchmarking to discover optimal system prompt configurations.

• 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.

Practitioners like @jerryjliu0 are highlighting that as model capabilities advance, the bottleneck shifts back to subtle issues in data quality that can poison agent generalization.

A discussion is surfacing around the critical need for high-quality data curation, specifically for filtering synthetic data used in agent training.

• Jerry Liu@jerryjliu0repeated

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