Skip to main content
ArcSight

← Homepage

Impact analysis example

What breaks when you change LangChain.js (libs/langchain-core)?

Real arc impact output on LangChain.js (libs/langchain-core) — same structure as the homepage demo. Know blast radius before you merge.

Project: LangChain.js (libs/langchain-core) · Score: 66.4/100 (Moderate technical debt) · 248 modules · 973 edges

Engineering decision simulation

Validation procedure 01 · merge vs. hold before output

Before arc impact

You plan a change to a high fan-in module in LangChain.js (libs/langchain-core). Would you merge without knowing blast radius?

Typical answer: hold — need dependency context.

After arc impact

Structured output below: score, hotspots, corridors, and high-risk modules — same field schema as the homepage proof.

Decision: confirm hold, narrow change, or inspect paths first.

Published as part of the validation record. Run locally: arc impact <file-you-are-changing>

Architectural Risk Map

Where risk accumulates by control surface — drill down Surface → Substrate → Mechanism → Evidence.

See control surfaces →

Dependency evidence

Architecture visualization

Diagrams are generated from ArcSight analysis data. Dependency graph and risk map show key modules; subsystem view shows high-level structure.

Dependency graph
Dependency graph for this codebase
Subsystem architecture
Subsystem architecture for this codebase
Risk map
Risk map for this codebase

Architecture score & metrics

Score: 66.4/100 · Stability: 10/100 · Status: at_risk · Confidence: 100%

  • Dependency cycles: 88
  • Layer violations: 3
  • Duplicate authorities: 49

Modules to inspect

One row per module. Fan-in is direct dependents. Blast radius is downstream modules reachable from that file.

Fan-in, direct dependents
  • src/messages/base24 dependents
  • src/messages/ai23 dependents
  • src/messages/content21 dependents
  • src/load/serializable15 dependents
  • src/messages/utils14 dependents
  • src/messages/block_translators12 dependents
  • src/messages/tool12 dependents
  • src/messages/block_translators/utils12 dependents
Blast radius, downstream modules
  • src/messages/content134 modules
  • src/load/serializable134 modules
  • src/messages/base131 modules
  • src/messages/ai131 modules
  • src/messages/utils131 modules
  • src/messages/block_translators131 modules
  • src/messages/tool131 modules
  • src/messages/message131 modules
ModuleFan-inBlastRisk
src/messages/content21134CRITICAL
src/load/serializable15134CRITICAL
src/messages/base24131CRITICAL
src/messages/ai23131CRITICAL
src/messages/utils14131CRITICAL
src/messages/block_translators12131CRITICAL
src/messages/tool12131CRITICAL
src/messages/message10131CRITICAL
src/errors6131CRITICAL
src/messages/block_translators/utils12CRITICAL

Architectural gravity centers

  • src/messages/content— concentrates too many responsibilities
  • src/messages/content— shared utility hub
  • src/load/serializable— concentrates too many responsibilities
  • src/load/serializable— shared utility hub
  • src/messages/base— concentrates too many responsibilities

CLI workflow

Scan → Fix → Finish → Watch. Nothing else required.

  • arc scanFind issues
  • arc fixApply fixes
  • arc finishResolve remaining
  • arc watchKeep clean

More commands (explain, impact, guard…) →

Key takeaways

  • •Structural scan reports dependency cycles — inspect impact paths before refactors that cross module boundaries.
  • •High fan-in modules are change-risk zones; run arc explain <path> before refactoring.
  • •Run the full workflow (analyze, hotspots, explain, impact, snapshot, drift) in your repo for deterministic architecture insights.

Run ArcSight on your repo

Install the CLI and run analyze, then hotspots and explain for the same workflow.

npm install -g arcsight
arc analyze
Install ArcSight →

This case study is generated from ArcSight analysis data. View short analysis. View all analyses