---

name: deps description: Visualize the project dependency graph and detect circular dependencies. Use this before refactoring, during architecture reviews, before adding new domains, or when evaluating module separation. Trigger immediately when cross-domain coupling seems high or circular references are suspected. Triggers on /deps, "dependency graph", or "check circular dependencies". version: 1.1.0 updated: 2026-02-11

Dependency Graph Analysis

Visualize dependency relationships across the project, detect circular dependencies, and analyze coupling between domains.

---

Skill Triggers

• /deps — Analyze all domain dependencies
• /deps <domain> — Dependencies for a specific domain
• /deps --cycles — Detect circular dependencies only
• /deps <file> --tree — Dependency tree for a file
• /deps --matrix — Cross-domain dependency matrix
• "show me the dependency graph"
• "check for circular dependencies"

---

Absolute Prohibitions

1. Do not modify code — analysis only.
2. Do not guess at dependencies — always parse actual import/require statements.
3. Do not ignore circular dependencies — report any that are found.

---

Execution Steps

Receive /deps [target] [options]
    |
    v
[1] Scope Resolution
    |
    v
[2] Domain Discovery
    |
    v
[3] Import/Require Parsing (Dependency Extraction)
    |
    v
[4] Graph Construction
    |
    v
[5] Cycle Detection
    |
    v
[6] Cross-Domain Coupling Analysis
    |
    v
[7] Output results

---

Step 1: Scope Resolution

Input	Scope
(none)	All project domains
Domain name	That domain's internal + external dependencies
File + --tree	File's dependency tree

Step 2: Domain Discovery

Domain identification criteria (in priority order):

1. domains definition in .claude/project-team.yaml
2. domains/<domain-name>/ directory
3. src/<domain-name>/ directory
4. Top-level meaningful directories

Layers vs. Domains:

• Layers (not domains): api/, services/, models/, utils/
• Domains (analysis targets): order/, member/, product/, payment/

Step 3: Import/Require Parsing

Language	Pattern
Python	from x.y import z, import x.y
JS/TS	import ... from '...', require('...')
Go	import "..."

# Python
grep -rn "^from \|^import " --include="*.py" <target-path>

# JS/TS
grep -rn "import.*from\|require(" --include="*.ts" --include="*.js" <target-path>

Step 4: Graph Construction

Node: Domain or module
Edge: Dependency relationship (directed)
  - from: the depending side
  - to:   the depended-on side
  - weight: number of imports

Step 5: Cycle Detection (DFS-based)

Level	Description	Severity
Domain level	A -> B -> A	CRITICAL
Module level	A -> B -> C -> A	HIGH
File level	A <-> B	MEDIUM

Step 6: Coupling Analysis

Metric	Description	Calculation
Ca (Afferent)	Incoming dependency count	Number of external domains that depend on this domain
Ce (Efferent)	Outgoing dependency count	Number of external domains this domain depends on
I (Instability)	Instability index	Ce / (Ca + Ce)

Grade	Cross-dependencies	Assessment
Loose	0–2	Healthy
Moderate	3–5	Manageable
Tight	6–10	Consider refactoring
Tangled	11+	Refactoring required

Step 7: Output

See references/output-formats.md for detailed output format.

---

Architecture File Generation

Analysis results are saved to .claude/architecture/dependencies/:

• domain-graph.mmd — Mermaid diagram
• module-graph.json — Module-level dependencies
• api-graph.json — API call relationships

---

Relationship with the Impact Skill

Aspect	/impact	/deps
Analysis unit	Single file	Domain / full project
Primary output	Impact report	Mermaid diagram
Core function	Change impact scope	Circular deps, coupling
Purpose	Pre-change analysis	Architecture review

---

Related Skill Integration

Skill	When to Link	Purpose
/impact <file>	When circular dependency is found	Check file-level impact
/changelog <domain>	When dependency changes occur	Recent change history
/architecture	After full analysis	Update architecture documents

---

Reference Documents

• references/output-formats.md — Output format details
• references/examples.md — Usage examples