Repository intelligence

hoangsonww/Claude-Code-Agent-Monitor

A real-time monitoring and analytics platform for visualizing Claude Code and Codex agent sessions, tool execution timelines, and subagent orchestration.

CLOUDM0N decision

REVIEW BEFORE ADOPTION

Trust REVIEW · 20/100

Good fit if

Developers utilizing Claude Code or Codex CLIs who need real-time, visual tracking of tool runs, task completion, and execution states

Watch out for

High-availability scaling patterns like horizontal pod autoscaling (HPA), active-active replicas, and blue-green/canary rollouts are unsupported to protect the single-writer SQLite database architecture

Practical intelligence

What matters before you adopt it

Decision first · detail on demand

Problem it solves

The lack of central visibility, real-time analytics, and operational control over local or remote command-line AI coding agent runs and their nested background subagents.

Best for

Developers utilizing Claude Code or Codex CLIs who need real-time, visual tracking of tool runs, task completion, and execution states

Teams looking to monitor local and remote multi-machine agent activity under a single, unified analytical interface

Integrators wanting to set up proactive alerts and webhook notifications on agent inactivity, errors, or token cost thresholds

Main trade-offs

Heuristic watchdog methods are required to recover and mark 'Waiting' on turns cancelled by pressing Escape, as the Claude Code CLI fails to fire a native hook upon user-interrupts

Self-updating and automated server restarts are explicitly omitted from the APIs by design to keep the platform compatible with diverse process supervisors

Why it stands out

D3.js-powered analytics containing 11 distinct visualization components, complete with localized and overflow-clamped tooltips

Graceful database fallback to Node's built-in 'node:sqlite' if native 'better-sqlite3' compilation fails on the host system

Intelligent process-liveness watchdog that automatically completes orphaned or dead CLI sessions left open during dashboard restarts

Trust & CVEs

Security evidence without the noise

Trust remains a decision signal; CVEs and scanner evidence explain what is driving the risk.

Security findings

Trivy + CLOUDM0N

Critical

High

Medium

Low

View trust evidence & security findings▼

Why this score

CLOUDM0NCLOUDM0N scanner found behavior that requires manual review.

TRIVYTrivy found 0 high vulnerabilities and 11 high/critical misconfigurations.

OPENSSFOpenSSF Scorecard essential security checks scored 5.0/10 (partial check set).

CLOUDM0N findings

HIGH

Remote download piped or chained into a shell requires manual review. 1 sample match(es) found.

MEDIUM

The project can spawn operating-system processes; review command construction and input handling. 5 sample match(es) found.

MEDIUM

2 lifecycle install script(s) require review.

Architecture from code16 modules · 3 edges▼

Structural evidence

Modules and dependency edges extracted from repository code. This is code evidence, not README inference.

Code files

370

Modules

Dependency edges

Core modules

(root)

1 files

bin

executables

1 files

client

142 files

desktop

15 files

mcp

48 files

monitoring

monitoring / alerting

7 files

scripts

14 files

server

135 files

statusline

1 files

vscode-extension

2 files

wiki

4 files

Dependency flow

scripts→server

server→monitoring

server→scripts

Detected languages

TypeScript · JavaScript · Shell · Python

Detected frameworks

Express

Architecture evidence details

flowchart TD
    %% agent-dashboard — high-level architecture (DRAFT, refine me)
    n0["(root) · 1 file"]
    n1["bin · executables · 1 file"]
    n2["client · 142 files"]
    n3["desktop · 15 files"]
    n4["mcp · 48 files"]
    n5["monitoring · monitoring / alerting · 7 files"]
    n6["scripts · scripts · 14 files"]
    n7["server · 135 files"]
    n8["statusline · 1 file"]
    n9["vscode-extension · 2 files"]
    n10["wiki · 4 files"]
    n6 --> n7
    n7 --> n5
    n7 --> n6
    class n1,n5,n6 infra
    classDef infra fill:#b35c00,color:#ffffff,stroke:#8f4a00

Evidence, security & integrations▼

Integrations

Claude CodeCodexCursorSlackDiscordMicrosoft TeamsGoogle ChatMattermost

Security notes

• The dashboard binds strictly to the 127.0.0.1 loopback interface by default; exposing it requires setting DASHBOARD_HOST and establishing a DASHBOARD_TOKEN

• All file paths are canonicalized via realpath during exploration, and symlinks attempting to escape trusted Claude directories are rejected

• Secrets, credentials, and API tokens are redacted dynamically before raw configuration files or settings are displayed

Still unknown

• The README does not provide concrete system requirements like minimum RAM or CPU specifications for executing high-frequency subagent streams

• No performance benchmarks are provided comparing database write latencies between node:sqlite and better-sqlite3

Adoption guidance▼

Adopt if

+ You run extensive, multi-hour local Claude Code or Codex sessions and need structured, visual feedback on subagent performance and cost

+ You must maintain an offline, air-gapped dev environment where telemetry and web assets cannot be retrieved from external hosts

+ You require custom, rule-based alerts dispatched to incident platforms like PagerDuty or Slack when automated runs fail

Avoid if

– You are running a high-availability server on Kubernetes and require active-active replicas, scale-up rules, or canary deployments

– You expect an automated, one-click button inside the UI to update the underlying system repository files and restart the web server

– You need native system-level process monitoring and automatic cleanup sweeps for CLI instances running on Windows machines

How it works & getting started▼

How it works

1.The Claude Code or Codex CLI fires an event-driven hook or appends to on-disk session transcripts

2.A local hook-handler script reads the event JSON from stdin and securely forwards the payload to the running Express server

3.The Express server writes events and token structures within SQLite database transactions and scans subagent-specific logs to attribute nested tools

4.The server publishes real-time state changes and progress data over a WebSocket connection to the front-end client

5.The React-driven browser client, Electron app, or VS Code extension immediately re-renders dashboard cards, Kanban lanes, and D3 graphs with zero polling

Getting started

→ Install project dependencies and build workspace tools by running: npm run setup

→ Initialize and register Claude Code or Codex hooks interactively with: npm run install-hooks

→ Launch the server and watch-mode Vite client simultaneously by running: npm run dev

→ Open the development client interface at: http://localhost:5173

Agent handoff

Use with any agent

JSON API

Alternatives

Nearby repositories worth comparing before adoption.

Compare top options →

velpos

Jxin-Cai/velpos

Fit

Rapidly build a multi-Agent management dashboard based on Claude Code, integrating daily work and remote IM collaboration. It includes Claude Code monitoring, channel configuration and message notifications, with a Codex GUI optimized for non-technical users.

Trust REVIEW · 65

Compare →

frieren-agent-monitor

Eskeeet/frieren-agent-monitor

Fit

A local open-source Frieren desktop companion for monitoring local and remote Pi, Codex, Claude Code and Cursor sessions.

Trust REVIEW · 90

Compare →

agentops

AgentOps-AI/agentops

Fit

Python SDK for AI agent monitoring, LLM cost tracking, benchmarking, and more. Integrates with most LLMs and agent frameworks including CrewAI, Agno, OpenAI Agents SDK, Langchain, Autogen, AG2, and CamelAI

Trust REVIEW · 80

Compare →