DepRank

PageRank your codebase — know which files are load-bearing before your AI agent rewrites them

npx -y github:Isco0819/dep-rank

Ever run into this?

Like madge (8k+ GitHub stars) and dependency-cruiser (3k+ stars), devs clearly want dependency graph tools — but neither ranks files by criticality. AI coding agents (Cursor, Claude Code) refactor blindly: one change to a high-fan-in file cascades 40 test failures. There is no fast way to ask 'how load-bearing is this file?'

The fix

dep-rank runs static import analysis on your TS/JS project and assigns every file a PageRank-style criticality score plus blast-radius count. An MCP server mode exposes a check_criticality tool so agents can query file risk before any edit.

What it does

static import graph builder (ESM/CJS/TS path aliases) — zero runtime required
PageRank score + blast-radius count per file, JSON or pretty-print CLI output
MCP server: check_criticality(path) returns score + top-5 downstream dependents so agents self-throttle on risky edits

npx -y github:Isco0819/dep-rank