ArcSight MCP Server
Analyzes JavaScript codebases to build dependency graphs, identify high-impact shared files, and compute downstream blast radius before edits.
Analyzes PHP codebases to build dependency graphs, identify high-impact shared files, and compute downstream blast radius before edits.
Analyzes Python codebases to build dependency graphs, identify high-impact shared files, and compute downstream blast radius before edits.
Analyzes TypeScript codebases to build dependency graphs, identify high-impact shared files, and compute downstream blast radius before edits.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@ArcSight MCP Server check the downstream impact before I edit src/auth/session.ts"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
ArcSight MCP — Stop Cursor & Claude from Breaking Working Code
ArcSight gives Cursor, Claude Code, and Windsurf deterministic x-ray vision into your codebase's dependency graph (TypeScript, JavaScript, Python, and PHP).
One command automatically configures your MCP server and agent guardrails so your AI checks downstream blast radius—and stops to ask confirmation—before editing high-impact shared files.
Quickstart (30 Seconds)
Run this inside the root of any Git repository:
npx arcsight initThen reload Cursor (or restart Claude Code once) so the arcsight MCP server activates.
What npx arcsight init Does
Auto-configures your AI editors: Writes
.cursor/mcp.json,.mcp.json(Claude Code), and.windsurf/mcp.jsonpointing tonpx -y arcsight mcp.Injects pre-edit guardrails: Adds
.cursor/rules/arcsight.mdcand a marked block inCLAUDE.mdinstructing the agent to runcheck_file_impactbefore editing shared files andarchitecture_statusbefore finishing.Reveals single points of failure: Immediately prints your top shared hubs in plain English (e.g.,
src/auth/session — 26 files depend on this. It sits on Login.).
Related MCP server: codememory
MCP Tools Included
check_file_impact: Computes direct callers and full downstream blast radius before an agent edits a file. Flags high-impact hubs (>20dependents) and instructs the agent to stop and confirm.architecture_status: Verifies that recent edits did not introduce new circular dependencies or structural regressions.
Privacy & Security
100% Local: All dependency parsing and graph analysis run locally on your machine. Your source code is never uploaded.
Free: All CLI and MCP commands are free to use via
npx arcsight init.
Server entrypoint
This repository does not contain the ArcSight engine. index.js starts the published arcsight@1.0.30 package over stdio:
npm install
node index.jsLinks
Website & Observatory Reports: https://arcsight.ai
npm Package: https://www.npmjs.com/package/arcsight
Issues & Feedback: Open an issue
Available Tools
4 toolsarchitecture_statusA
Session map: primary language, top hubs, open boundaries, and the unresolved file count. Call this when impact hooks did not run or the caller set is unknown. Writes nothing.
Inputs: repo.
check_diff_impactC
Combined blast radius and the direct callers of the changed files. Read decision first. high means the count is over max. Pass includeAffected only for the full module list.
Inputs: paths (required), max, repo, files, +1 more.
check_file_impactC
Hub size and the direct callers of one file. Read decision first. high means the count is over max. Pass includeAffected only for the full module list.
Inputs: path (required), max, file, repo, +1 more.
plan_fixC
Call only when the user asked for a mechanical fix of a cycle, a layer edge, or an unused import. Writes nothing. Skip for a behavior change. Pass confirm mechanical.
Inputs: repo, confirm.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 4 tools
check_file_impact and check_diff_impact both analyze impact but are distinguished by scope (single file vs. changed files), and their descriptions clarify this. plan_fix and architecture_status serve clearly different purposes, though the two impact tools could still be momentarily confused.
Three tools follow a verb_noun pattern (check_file_impact, check_diff_impact, plan_fix) in snake_case, while architecture_status is a noun phrase. This is a minor deviation but still readable and consistent in casing.
Four tools are reasonable for a focused architecture impact analysis server, covering the core read-only operations without bloat. It sits at the lower end but each tool has a distinct role.
The set covers file impact, diff impact, architecture status, and mechanical fix planning, which are the core workflows. Minor gaps exist—no tool to apply a fix or inspect a specific dependency graph—but agents can work around these for the stated advisory purpose.
Maintenance
Related MCP Connectors
Hosted code graph over MCP: exact callers, dependencies, and cross-repo blast radius for AI agents.
IaC, CI and dependency security review for coding agents: local scan CLI plus hosted OAuth MCP.
Codebase knowledge graph — architectural awareness for AI coding agents via MCP
Free MCP security tools for AI coding agents. Check npm/PyPI package versions, identify known OSV vulnerabilities, compare with the latest releases, and review up to 3 dependencies before deployment.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables AI coding agents to query a codebase as a knowledge graph, providing token-budgeted context, search, and impact analysis via MCP tools.27 PyPIMIT
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to locally search, query, and understand codebases with token-efficient context, dependency graphs, history, architecture diagrams, and metrics through MCP.15 npm3MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI coding agents to query a local, continuously updated symbol graph of a codebase, providing ranked search, caller/callee exploration, dependency paths, and git-diff impact analysis through MCP tools.14 npmApache 2.0
- AlicenseBqualityBmaintenanceEnables AI coding agents to compile repositories into high-signal, cache-pinned context and to query dependency centrality, blast radius, AST skeletons, circular dependencies, hotspots, and auto-generated agent rules through MCP tools.71MIT