Claude Code Found Silently Marking API Requests via Prompt Steganography
A security analysis of Anthropic’s Claude Code tool reveals hidden system prompt modifications that flag requests routed through specific domains and timezones.
A security analysis of Anthropic’s Claude Code tool reveals hidden system prompt modifications that flag requests routed through specific domains and timezones.
A default cleanup setting in Anthropic’s Claude Code command-line tool silently purges project conversation logs older than 30 days without warning.
ZLUDA Version 6 is out, bringing basic texture support for Blender, legacy 32-bit PhysX for AMD hardware, and streamlined Windows setup.
Robusta CEO Natan Yellin warns that manually feeding terminal errors to Claude Code breaks the agentic loop, urging developers to grant tool access instead.
A new open-source standard aims to solve the AI memory silo problem, letting users share context across ChatGPT, Claude, and developer tools.
Workweave’s new open-source router acts as a local proxy for Anthropic, OpenAI, and Gemini, promising 50ms routing and up to 70% cost savings.
Inkeep has released OpenKnowledge, a local-first markdown editor and LLM wiki integrating Claude, Codex, and Cursor under an open-source license.
RubyLLM provides a single, standardized interface for Ruby developers to integrate major AI providers, streamlining workflow across different models.
Adobe has launched a public beta for conversational AI assistants across its Creative Cloud suite, including Photoshop and Premiere Pro, to automate tasks and streamline creative workflows.