Strukto Launches Mirage: Unified Virtual Filesystem for AI Agents Across S3, Drive, GitHub, and More
Strukto launched Mirage, a unified virtual filesystem that mounts disparate data sources — S3, Google Drive, Slack, Gmail, GitHub, Linear, Notion, Postgres, MongoDB, and more — side-by-side with standard Unix semantics (cat, grep, head, wc work natively across all). Built in six weeks at 1.1M+ lines of code, Mirage supports versioned workspace snapshots with one-call rollbacks, a two-layer cache for fast agent loops, and integrates with OpenAI Agents SDK, LangChain, Claude Code, and Vercel AI SDK.
Why It Matters
Mirage addresses one of the largest friction points in production agent deployment: agents that need to work across heterogeneous enterprise data sources without bespoke integrations for each. Filesystem abstraction as a universal interface could simplify agent harness architecture significantly.