Embedded Device Logger

The Embedded Device Logger is a Visual Studio Code extension that can connect to your devices over SSH, tail their logs, and help you analyze the data with loglevel colorization, quick filters, custom keywords highlights and filtered export. It provides also an SFTP client, SSH terminals and one-off SSH commands to help you develop, debug and maintain your Linux-based devices.

• Live logs view:

• SFTP Panel view and SSH terminal:

• Offline logs view:

If you like the extension, please rate it. We welcome issue reports and feature requests.

Key Features

• Stream device logs over SSH with real-time log-level parsing and colorization.
• Search, filter, bookmark, and export the exact lines you need.
• Highlight up to 10 keywords per panel to spot critical events fast.
• Run one-off SSH commands and optionally upload a script to /tmp, chmod 777 it, and execute it after the command.
• Open SSH terminals, including command panels that can optionally re-run after reconnecting and run uploaded scripts.
• Open device URLs in an external browser or in VS Code's embedded browser.
• Optionally ping all configured devices, with manual or scheduled checks and reachability dots in the Devices view.
• Browse files with the built-in SFTP explorer, including quick search, find/grep result views, and keyboard shortcuts.
• Assign per-device colors to make tabs and device lists easier to scan.
• Organize devices with collapsible groups in the Embedded Devices view.
• Choose the extension UI language from English, Spanish, Italian, Simplified Chinese, Traditional Chinese, French, German, Japanese, Korean, Russian, Portuguese (Brazil), Turkish, Polish, Czech, or Hungarian, following VS Code's display language by default and falling back to English for missing strings.
• Secure by default: passwords and key passphrases live in VS Code Secret Storage.
• Privacy focused. No telemetry. Everything runs locally.

Getting started

1. Install the extension (see below).
2. Open the Embedded Logger view from the Activity Bar (terminal icon).
3. Open the configuration with the edit icon (🖍) to launch the Device Manager, add your devices and start streaming logs.

For the full setup and configuration reference, see the Detailed Usage and Configuration guide.

Installation

• From the VS Code Extensions view, search for Embedded Device Logger (Publisher: Scallant),
• or from Quick Open (Ctrl/Cmd+P): ext install Scallant.embedded-device-logger,
• or from a terminal: code --install-extension Scallant.embedded-device-logger.

Visit the Marketplace page for more details.

Motivation behind the development of this VSCode Extension

When you develop, debug, or audit software for embedded Linux devices, logs are everything.

They tell you what happened, when it happened, and often why it happened.

Yet in practice, working with logs on embedded systems is still surprisingly awkward.

Most of us rely on:

• SSH into the device
• Running tail -f, journalctl, or custom scripts
• Copy-pasting outputs
• Repeating the same commands again and again

And while VS Code has become the de-facto development environment for many engineers, log inspection still lives mostly outside the editor.

I tried to find a VS Code extension that was:

• Fast
• Simple
• Designed for embedded Linux, not servers
• Capable of real-time and offline log analysis

I couldn’t find one that fully fit that workflow.

So I built it.

More about this story at Medium Article

For developers

Want to build from source or contribute? npm run compile type-checks the extension and creates a readable bundle at out/extension.js, while npm run watch keeps that output up to date during development. Packaging automatically runs npm run compile:production to minify the bundle; source maps and documentation screenshots are excluded from the VSIX to reduce download size. See the Developer Setup and Workflow for packaging, local installs, and contribution guidelines. The project is open to pull requests. Please, check the CONTRIBUTING guide and the Code Architecture Overview before submitting.

AI / MCP integration

Embedded Device Logger includes a disabled-by-default local MCP server for coding and debugging agents. It uses MCP Streamable HTTP at http://127.0.0.1:39070/mcp because an external client can attach to the running extension, which owns device authentication. The port is configurable, but the server always binds to 127.0.0.1; there is no LAN bind setting.

MCP clients do not receive device SSH credentials. Device authentication remains managed by Embedded Device Logger and its existing secure credential storage.

MCP does not provide arbitrary remote shell access. Device-changing operations are limited to existing custom commands explicitly authorized by the user for MCP access.

Configuration and connection

Use Embedded Logger: Enable MCP Server, Disable MCP Server, Show MCP Status, Show MCP Connection Configuration, and Show MCP Task Logs. The configuration command displays and copies this client configuration:

{"servers":{"embeddedLogger":{"type":"http","url":"http://127.0.0.1:39070/mcp"}}}

Capabilities and security

Log/data tools are list_devices, get_device_status, get_recent_logs, search_logs, get_logs_around, and list_log_sources. They read immutable, bounded snapshots without modifying UI filters. Diagnostic tools are get_system_info, get_service_status, get_service_logs, get_processes, get_disk_usage, get_memory_info, get_network_info, and get_device_uptime. They use predetermined, read-only Linux commands through the extension's existing SSH/authentication infrastructure; validated parameters cannot supply shell syntax. Devices must already have an active connected log panel.

Authorized-action tools are list_custom_commands and run_custom_command. Enable embeddedLogger.mcp.allowCustomCommands and the command's Allow MCP checkbox. Existing commands default to denied. mcpConfirmation defaults to always, causing a modal VS Code confirmation; dismissal denies execution. The agent receives an opaque command ID and name, never its shell text, and cannot provide arguments, environment, working directory, hosts, or credentials.

All MCP output and safe errors pass through centralized redaction and size limits; original UI/saved logs are unchanged. Requests, logs, searches, journal lines, diagnostic execution time, and outputs are bounded. Secret Storage is not reachable. Device metadata omits hosts, usernames, key paths, fingerprints, passwords, passphrases, and bastions. Custom actions write metadata-only audit events (timestamp, device, command, success, duration) to the MCP Output Channel without stdout/stderr or secrets. The listener and sockets are unreferenced so MCP cannot keep VS Code alive.

Current limitations: live MCP history starts when a remote log panel collects lines; imported offline files remain UI-only. Diagnostics require common Linux utilities such as systemd/journalctl, ip, ss, and /proc. Diagnostic execution reuses the existing bounded SSH command runner rather than exposing the underlying connection or credentials.