Denoise · Twitter

The security focus is shifting from prompt injection in models to vulnerabilities in the agent's orchestration and tool-use layer.

Major labs like Anthropic and Google DeepMind are releasing detailed write-ups and frameworks for red-teaming autonomous agents.

• 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 primary developer interface is shifting from IDE plugins like Copilot to standalone terminal agents, a trend confirmed by @karpathy and early adopters like @levelsio.

Anthropic released Claude Code 1.5, a terminal-native agent, prompting benchmarks and discussion about moving away from 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 complete stack for agent deployment is emerging, with OpenAI providing orchestration primitives, Vercel offering edge runtimes, and LangChain ensuring protocol interoperability.

OpenAI, Vercel, and Replit released new SDKs, runtimes, and deployment tools specifically for building and hosting autonomous agents.

• 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.

Foundational dataset releases remain a key strategy for players like MistralAI to enable open community development in the multimodal space.

MistralAI released a large, cleaned web OCR dataset to facilitate the training of new multimodal models.

• Mistral AI@MistralAIrising

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

As context windows grow to millions of tokens, the conversation is shifting from simple vector retrieval (RAG) to more sophisticated memory management, as articulated by @GregKamradt and @mem0ai.

Discussions focus on the limitations of RAG in large-context models, proposing new frameworks for 'context engineering' and structured 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.

Automation is becoming a standard feature in SaaS. Notion and Linear are building task-specific agents, while platforms like Temporal provide the generalized, durable primitives needed for more complex agent workflows.

Workspace tools like Notion and Linear are shipping autonomous features, while infrastructure providers like Temporal highlight their role in agent orchestration.

• 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 process of finding effective system prompts is becoming a data science problem, with firms like Weights & Biases running massive benchmarks to find the efficient frontier.

Practitioners are sharing systematic approaches to prompt optimization, moving from anecdotal tricks to large-scale benchmark-driven engineering.

• 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.

For agent training, data quality is paramount. The key challenge is not generating synthetic data, but curating it to ensure it improves, rather than poisons, model performance.

A post from @jerryjliu0 highlights the difficulty of filtering synthetic data to avoid negatively impacting an agent's generalization capabilities.

• Jerry Liu@jerryjliu0repeated

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