Denoise · Twitter

The security focus is shifting from prompt injection in single models (Anthropic) to vulnerabilities in the multi-agent orchestration layer (GoogleDeepMind, @AlexAlbert__).

Major labs are publicly disclosing agent jailbreaks and releasing formal red-teaming frameworks, while independent researchers test agents in live pentesting scenarios.

• 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.

Competition is intensifying between agent-native workflows (Anthropic) and IDE-integrated assistants (Copilot), with early benchmarks from @swyx suggesting performance divergence.

Anthropic's release of Claude Code 1.5, a terminal-native agent, is driving a conversation about a fundamental 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.

A convergence is visible around providing managed, serverless infrastructure for 'agent workers,' with OpenAI, Vercel, and Replit all building primitives for durable, multi-agent execution.

Major infrastructure providers including OpenAI, Vercel, and Replit are releasing SDKs and runtimes to standardize the deployment and orchestration of 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.

While the agent ecosystem expands rapidly, foundational dataset releases like MistralAI's remain a key, though less frequent, driver of progress for training base models.

MistralAI released a large-scale, cleaned web OCR dataset for public use.

• Mistral AI@MistralAIrising

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

With 10M context windows tested (@reach_vb), the tension is between 'stuffing the context' and using structured systems like knowledge graphs (@llamaindex) or specialized memory layers (@mem0ai).

The discourse around context is evolving from simple RAG to 'context engineering,' managing large memory windows and implementing more complex retrieval and caching strategies.

• 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.

SaaS platforms like Notion and Linear are integrating autonomous workflow features, suggesting a convergence where enterprise tools and AI agents are solving similar orchestration problems.

Workspace automation tools like Notion and Linear are releasing features that auto-fill, auto-triage, and chain updates, mirroring agent-like capabilities in a SaaS context.

• 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 bifurcating between artisanal prompt crafting (@dotey) and systematic, large-scale optimization of system prompts (Weights & Biases, DSPy).

Practitioners are sharing specific techniques for production prompts, while organizations like Weights & Biases are benchmarking system prompt variations at scale.

• 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 grow, the bottleneck shifts back to high-quality data. The focus is now on subtle data poisoning issues in synthetic datasets for agent tuning.

Discussion from @jerryjliu0 highlights the challenges of dataset curation for agent training, specifically filtering synthetic data that harms generalization.

• Jerry Liu@jerryjliu0repeated

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