Archify
A repository-understanding and documentation system that extracts grounded architecture and codebase context so developers can study projects faster and generate consistent technical documents from real evidence.
Archify is a repository-understanding and documentation system that turns an existing codebase into grounded architecture knowledge instead of forcing developers to reconstruct that knowledge by hand every time they need to study, explain, or extend a system.
The workflow is layered, but the user experience is simple: run `npx archify init` once, then invoke the Archify skill whenever repository understanding or a document is needed. The skill checks status, refreshes `.archify` evidence when needed, and runs the internal analysis and generation flow automatically.
Inspiration Behind It
The project came from the repeated cost of understanding existing repositories well enough to document them or build similar systems. That work usually means slow manual reconstruction across code, services, APIs, data models, and scattered docs.
Archify was built around the idea that repository understanding is the real bottleneck. If the evidence layer is weak, every downstream summary, architecture document, or generated guide becomes weaker too.
System Architecture
Architecture Diagram
Archify
9 nodes
About The Flow / Architecture
Archify combines a Python-based analysis engine, AST-driven extraction, JavaScript package and setup flow, a generated `.archify` evidence directory, and a synthesis layer that creates artifacts such as `design-packet.json` and document-specific guides. That separation matters because setup, evidence extraction, and document generation are different responsibilities and should not collapse into one opaque workflow.
After one-time setup, the skill becomes the main interface. On invocation, it checks whether repository evidence is current, runs analysis if `.archify` knowledge is missing or stale, prepares synthesis artifacts and guides, and then creates the requested output with the right guide. Each guide defines which artifacts matter for a document, what order they should be read in, what can be stated as confirmed fact versus inference, how missing evidence should be handled, and how the final output should be validated.
Archify extracts grounded architecture and codebase context from a repository so it is easier to study, understand, and build similar systems. The user flow stays simple: run npx archify initonce, then use the Archify skill whenever repository understanding or a document is needed.
The core value is the grounded .archify evidence layer and the guides generated from it. The skill handles evidence refresh, grounded packet preparation, and guide-based document creation under the hood while keeping clearer boundaries between confirmed fact, inference, fallback reads, and validation.