Files over tools: how we built our agent with a virtual filesystem and bash
The engineering team at Knock explains how they transitioned from a tool-per-resource architecture to a virtual filesystem and bash-based approach for their AI agent. This shift allows the agent to manage complex customer messaging resources more efficiently by treating API interactions like file operations.
Why it matters
This represents a growing trend in AI engineering where developers are moving away from rigid tool-calling patterns toward more flexible, file-system-like interfaces for LLMs.
Claude Perplexity Google Gemini GitHub Copilot Grok In March 2026, we shipped the Knock Agent : an AI agent for managing all your customer messaging resources in Knock. It can create and update workflows, templates, and audiences, help segment users, and answer questions about how your customer messaging is performing. The agent can be invoked from the Knock dashboard, a connected Slack workspace, the API, or our MCP server.
Technical documentation focused on software architecture with no political or social agenda.
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in