impact-graph

"If I change this function, what breaks?"
Change impact analysis powered by call graphs + PageRank.

What it does

impact-graph statically analyzes your TypeScript/JavaScript and Python codebase to build a call graph. It then answers: "if I refactor this function, what else could break?"

• 📊 Call graph construction — extracts function defs, calls, imports, class methods
• 🎯 Impact analysis — BFS traversal to find all transitively affected functions
• 🔥 Risk scoring — PageRank-based risk: highly-depended-on functions = highest risk
• 🔄 Circular dep detection — Tarjan's SCC algorithm
• 🌑 Orphan detection — functions defined but never called
• 🔀 Git diff mode — automatically analyze what changed in your current branch

Install

npm install -g @phoenixaihub/impact-graph
# or
npx @phoenixaihub/impact-graph <command>

Usage

1. Index your project

cd my-project
impact-graph index

🔍 Indexing project at /my-project ...
✅ Indexed in 312ms
   Files:     47
   Functions: 284
   Calls:     891
   Graph saved to .impact-graph/graph.json

2. Check impact of a function change

impact-graph check src/utils.ts:processPayment

⚡ Impact Analysis — 12 affected function(s)

┌────────────────────────────┬──────────────────────────────┬───────┬────────────┬─────────┐
│ Function                   │ File                         │ Depth │ Risk Score │ Callers │
├────────────────────────────┼──────────────────────────────┼───────┼────────────┼─────────┤
│ handleCheckout             │ src/checkout.ts              │ 1     │ 12.45      │ 3       │
│ processOrder               │ src/orders.ts                │ 2     │ 9.12       │ 2       │
│ sendConfirmationEmail      │ src/notifications.ts         │ 3     │ 4.33       │ 1       │
└────────────────────────────┴──────────────────────────────┴───────┴────────────┴─────────┘

Flags:

• --depth <n> — max traversal depth (default: 10)
• --format json|table|markdown — output format

3. Analyze your git diff

impact-graph check --diff

Automatically detects changed functions in staged + unstaged git diff and runs impact analysis on each.

4. Visualize the graph

impact-graph visualize --format mermaid
impact-graph visualize --format dot > graph.dot && dot -Tsvg graph.dot -o graph.svg

Mermaid output (paste into GitHub/Notion):

graph LR
  add["add"]
  multiply["multiply"]
  multiply --> add
  square["square"]
  square --> multiply

5. Codebase health stats

impact-graph stats

📊 Codebase Health Stats

  Functions: 284
  Calls:     891
  Files:     47

🔥 Top 10 Riskiest Functions (PageRank)

  processPayment                 14.23 risk    12 callers  src/payments/processor.ts
  validateUser                   11.87 risk     9 callers  src/auth/validate.ts
  ...

🔄 Circular Dependencies: 2
  ⚠ processOrder → validateCart → processOrder
  ⚠ formatDate → parseDate → formatDate

🌑 Orphan Functions (never called, never call): 7
  legacyExportCSV  (src/export.ts:142)
  ...

🔗 Most Coupled Modules (by outgoing calls)
  src/checkout.ts          34 calls out

Supported Languages

| Language | Function defs | Calls | Imports | |----------|--------------|-------|---------| | TypeScript | ✅ | ✅ | ✅ | | JavaScript | ✅ | ✅ | ✅ | | Python | ✅ | ✅ | ✅ |

Go support coming soon.

Algorithms

• Call graph construction — regex-based AST extraction (tree-sitter optional)
• PageRank — iterative PageRank (85% damping, 50 iterations). High rank = many callers = risky to change
• Transitive closure — BFS from changed node following reverse call edges
• Circular dep detection — Tarjan's SCC algorithm
• Risk scoring — PageRank × 100 + depth_factor × 10

Development

git clone https://github.com/phoenix-assistant/impact-graph
cd impact-graph
npm install --legacy-peer-deps
npm run build
npm test

License

MIT © PhoenixAI Hub

Architecture

impact-graph
├── src/
│   ├── cli.ts         # Commander CLI entry point
│   ├── parser.ts      # AST extraction (TypeScript, JavaScript, Python)
│   ├── graph.ts       # Call graph construction + adjacency representation
│   ├── pagerank.ts    # Iterative PageRank + risk scoring
│   ├── impact.ts      # BFS transitive impact traversal
│   ├── stats.ts       # Codebase health metrics
│   └── visualize.ts   # Mermaid / DOT / JSON output
└── tests/

Flow: parse files → build call graph → (on query) BFS from changed nodes → rank by PageRank → output

CI Setup

Add impact analysis to your PR workflow:

# .github/workflows/impact.yml
name: Impact Analysis
on: [pull_request]

jobs:
  impact:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0
      - uses: actions/setup-node@v4
        with:
          node-version: 20
      - run: npm ci
      - name: Build impact graph
        run: npx @phoenixaihub/impact-graph index
      - name: Analyze PR impact
        run: npx @phoenixaihub/impact-graph check --diff --format markdown >> $GITHUB_STEP_SUMMARY

Contributing

See CONTRIBUTING.md.