9.5 KiB
Dependency Graph
ComfyFlow Insight
Technical Documentation
Overview
The Dependency Graph Engine is one of the most important systems inside ComfyFlow Insight.
Its primary responsibility is to understand how nodes depend on one another and determine the impact of modifying, disconnecting, or removing nodes from a workflow.
While the Branch Analyzer focuses on workflow splitting, the Dependency Graph focuses on workflow relationships.
It answers questions such as:
- Which nodes depend on this node?
- Which nodes does this node depend on?
- Is this node safe to remove?
- What will break if this node is removed?
- Which nodes are isolated?
- Which nodes are critical?
Why Dependency Analysis Matters
Large workflows often contain hundreds of node relationships.
Example:
CheckpointLoader
↓
KSampler
↓
VAE Decode
↓
Upscaler
↓
SaveImage
Removing:
VAE Decode
could break:
Upscaler
SaveImage
Without dependency analysis, it becomes extremely difficult to predict workflow impact.
Dependency Graph Objectives
The Dependency Graph Engine performs the following tasks:
Build Node Relationships
Detect Dependencies
Detect Dependents
Detect Isolated Nodes
Identify Critical Nodes
Measure Dependency Complexity
Support Impact Analysis
Support Workflow Repair
Support Workflow Advisor
Core Concepts
The Dependency Graph treats a workflow as a directed graph.
Example:
Node A
↓
Node B
↓
Node C
Relationship:
Node B depends on Node A
Node C depends on Node B
Dependency
Definition
A Dependency is a node required before another node can execute.
Example:
LoadImage
↓
ImageScale
Dependency:
ImageScale depends on LoadImage
Why Dependencies Matter
Dependencies determine:
Execution Order
Workflow Stability
Repair Risks
Failure Propagation
Dependents
Definition
A Dependent is a node that relies on another node.
Example:
LoadImage
↓
ImageScale
Dependent:
ImageScale
Why Dependents Matter
Dependents reveal:
Impact Scope
Workflow Reach
Critical Node Importance
Nodes with many dependents are often critical workflow components.
Example Dependency Graph
Workflow:
CheckpointLoader
↓
KSampler
↓
VAEDecode
↓
SaveImage
Dependencies:
KSampler
depends on
CheckpointLoader
VAEDecode
depends on
KSampler
SaveImage
depends on
VAEDecode
Dependency Count
Definition
Dependency Count measures how many upstream nodes are required.
Example
SaveImage
Depends on:
CheckpointLoader
KSampler
VAEDecode
Dependency Count:
3
Interpretation
0 Independent
1–3 Low
4–8 Moderate
8+ High
Dependent Count
Definition
Dependent Count measures how many downstream nodes rely on a node.
Example
KSampler
Feeds:
VAEDecode
PreviewImage
SaveImage
Dependent Count:
3
Why It Matters
Higher dependent counts indicate:
Critical Workflow Components
Higher Impact Risk
More Repair Sensitivity
Isolated Nodes
Definition
An Isolated Node has no workflow connections.
Example:
WorkflowInspector
No incoming links.
No outgoing links.
Example
LoadImage → SaveImage
PreviewImage
Node:
PreviewImage
is isolated.
Why Isolated Nodes Matter
Isolated nodes often indicate:
Unused Experiments
Development Artifacts
Workflow Mistakes
Forgotten Components
Safe Removal Detection
One of the primary responsibilities of the Dependency Graph Engine is identifying nodes that may be safe to remove.
Safe Candidate
Example:
PreviewImage
Conditions:
No Dependencies
No Dependents
No Workflow Impact
Result:
Potentially Safe To Remove
Important Warning
Safe removal suggestions are recommendations only.
Users must verify workflow intent before deleting nodes.
A node may be intentionally disconnected.
Critical Nodes
Definition
Critical Nodes are nodes with significant workflow influence.
Characteristics:
Many Dependents
High Dependency Traffic
Core Workflow Position
Example
CheckpointLoader
Feeds:
KSampler A
KSampler B
KSampler C
KSampler D
Impact:
Removing this node breaks
multiple workflow sections.
Dependency Chains
Definition
A Dependency Chain is a sequence of connected dependencies.
Example:
CheckpointLoader
↓
KSampler
↓
VAEDecode
↓
Upscaler
↓
SaveImage
Chain Length:
5
Why Chains Matter
Long chains increase:
Complexity
Maintenance Cost
Debugging Difficulty
Repair Risk
Dependency Complexity
Definition
Dependency Complexity measures how difficult workflow relationships are to understand and maintain.
Factors
The engine evaluates:
Dependency Count
Dependent Count
Chain Length
Branching Impact
Workflow Depth
Complexity Levels
Low
Characteristics:
Few Relationships
Short Chains
Easy Navigation
Medium
Characteristics:
Moderate Relationships
Several Branches
Manageable Complexity
High
Characteristics:
Large Dependency Chains
Multiple Critical Nodes
Complex Workflow Structure
Dependency Risk Analysis
The Dependency Graph Engine identifies risk levels.
Low Risk Node
Characteristics:
Few Dependents
Minimal Connections
Limited Impact
Medium Risk Node
Characteristics:
Moderate Workflow Influence
Several Dependents
High Risk Node
Characteristics:
Core Workflow Component
Many Dependents
Multiple Branches
Impact Analysis Support
The Dependency Graph Engine provides data to the Impact Analysis Engine.
Example:
Node:
KSampler
Dependents:
VAEDecode
PreviewImage
SaveImage
Impact:
Removing KSampler
affects 3 downstream nodes.
Workflow Repair Integration
Workflow Repair uses dependency information before generating recommendations.
Example:
Isolated Node Found
Dependency Graph Result:
Dependencies = 0
Dependents = 0
Repair Recommendation:
Review For Removal
Workflow Advisor Integration
Workflow Advisor uses dependency information when calculating workflow quality.
Example:
High Dependency Complexity
Advisor Recommendation:
Review Workflow Structure
Reduce Dependency Chains
Simplify Workflow Design
Common Dependency Patterns
Linear Pipeline
Example:
Load
↓
Process
↓
Save
Characteristics:
Simple
Easy To Maintain
Low Risk
Branching Pipeline
Example:
Input
│
├── Branch A
├── Branch B
└── Branch C
Characteristics:
Moderate Complexity
Higher Maintenance
Shared Dependency
Example:
CheckpointLoader
│
┌────┼────┐
▼ ▼ ▼
A B C
Characteristics:
Critical Node
High Impact Risk
Engineering Insights
Typical Dependency Insights:
Workflow contains isolated nodes.
Dependency complexity is moderate.
Critical workflow components detected.
Long dependency chains identified.
Workflow structure is highly interconnected.
These insights help users understand workflow behavior beyond simple node counts.
Example Analysis Report
Dependency Graph Report
Total Dependencies : 184
Critical Nodes : 12
Isolated Nodes : 3
Longest Dependency Chain : 18
Dependency Complexity : High
Interpretation:
Workflow contains
significant inter-node
relationships.
Impact analysis and
repair review are
recommended.
Best Practices
Review Isolated Nodes
Always verify whether isolated nodes are:
Unused
Experimental
Intentionally Disconnected
before removal.
Monitor Critical Nodes
Nodes with many dependents should be modified carefully.
Changes may affect large workflow sections.
Reduce Excessive Chains
Long dependency chains increase:
Maintenance Cost
Debugging Time
Repair Difficulty
Simplify where possible.
Relationship With Other Engines
The Dependency Graph Engine supports:
Branch Analyzer
Impact Analysis
Workflow Advisor
Workflow Repair
Workflow Actions
It acts as the central relationship engine inside ComfyFlow Insight.
Summary
The Dependency Graph Engine provides deep visibility into workflow relationships.
It identifies:
Dependencies
Dependents
Critical Nodes
Isolated Nodes
Dependency Chains
Dependency Complexity
Impact Risks
These insights help users safely modify workflows, understand workflow architecture, and make informed repair decisions.
Dependency analysis is the foundation for Impact Analysis, Workflow Advisor, and Workflow Repair recommendations.
Next Document
Continue with:
docs/workflow-advisor.md
This document explains workflow scoring, grading, production-readiness evaluation, risk assessment, recommendation generation, and engineering decision support.