dev-machine-guard
Scan your dev machine for AI agents, MCP servers, IDE extensions, and suspicious packages - in seconds.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Check it deeper on your machine
It has changed since its scan on 11 Nov 2026.
As a Claude Code plugin, install it and then ask Claude Code:
/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/step-security/dev-machine-guard with scanskillsafety
Or from a terminal, clone it and run it:
git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/step-security/dev-machine-guard'
It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.
Badge for your README
Markdown
[](/listings/step-security--dev-machine-guard)HTML
<a href="/listings/step-security--dev-machine-guard"><img src="/api/listings/step-security--dev-machine-guard/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
The steps below install the binary directly and are intended for **community users** evaluating Dev Machine Guard on a single machine.
**Enterprise customers should not install the binary manually.** Deploy Dev Machine Guard across your fleet using the loader script through your MDM or EDR tooling. See the [Installation Script documentation](https://docs.stepsecurity.io/developer-machines/installation-script) for the supported, auto-updating deployment flow.
Install from release (community)
Download the latest binary for your platform from [GitHub Releases](https://github.com/step-security/dev-machine-guard/releases). Release asset filenames embed the version (for example, `stepsecurity-dev-machine-guard-1.17.0-darwin`), so the snippets below resolve the latest tag first rather than hardcoding a version.
**macOS** (universal binary — Apple Silicon and Intel)
VERSION=$(curl -fsSL https://api.github.com/repos/step-security/dev-machine-guard/releases/latest | grep '"tag_name"' | cut -d'"' -f4 | sed 's/^v//')
curl -fsSL "https://github.com/step-security/dev-machine-guard/releases/download/v${VERSION}/stepsecurity-dev-machine-guard-${VERSION}-darwin" -o stepsecurity-dev-machine-guard
chmod +x stepsecurity-dev-machine-guard
# Run the scan
./stepsecurity-dev-machine-guard**Windows** (PowerShell)
$version = (Invoke-RestMethod https://api.github.com/repos/step-security/dev-machine-guard/releases/latest).tag_name.TrimStart('v')
$arch = if ($env:PROCESSOR_ARCHITECTURE -eq 'ARM64') { 'arm64' } else { 'amd64' }
Invoke-WebRequest -Uri "https://github.com/step-security/dev-machine-guard/releases/download/v$version/stepsecurity-dev-machine-guard-$version-windows_$arch.exe" -OutFile "stepsecurity-dev-machine-guard.exe"
# Run the scan
.\stepsecurity-dev-machine-guard.exe**Linux**
VERSION=$(curl -fsSL https://api.github.com/repos/step-security/dev-machine-guard/releases/latest | grep '"tag_name"' | cut -d'"' -f4 | sed 's/^v//')
ARCH=$(uname -m | sed 's/x86_64/amd64/;s/aarch64/arm64/')
curl -fsSL "https://github.com/step-security/dev-machine-guard/releases/download/v${VERSION}/stepsecurity-dev-machine-guard-${VERSION}-linux_${ARCH}" -o stepsecurity-dev-machine-guard
chmod +x stepsecurity-dev-machine-guard
# Run the scan
./stepsecurity-dev-machine-guardPre-built `.deb` and `.rpm` packages are also available on the [releases page](https://github.com/step-security/dev-machine-guard/releases).
Build from source
git clone https://github.com/step-security/dev-machine-guard.git cd dev-machine-guard make build ./stepsecurity-dev-machine-guard
Requires Go 1.26+. The binary has zero external dependencies.
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 626 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- <h1 align="center">StepSecurity Dev Machine Guard</h1>
- <p align="center">
- <img src="images/banner.png" alt="StepSecurity Dev Machine Guard — shield logo with terminal prompt" width="800">
- </p>
- <p align="center">
- <img src="images/demo.gif" alt="StepSecurity Dev Machine Guard demo" width="800">
- </p>
- <p align="center">
- <a href="https://github.com/step-security/dev-machine-guard/actions/workflows/tests.yml"><img src="https://github.com/step-security/dev-machine-guard/actions/workflows/tests.yml/badge.svg?branch=main" alt="Tests"></a>
- <a href="https://github.com/step-security/dev-machine-guard/actions/workflows/gosec.yml"><img src="https://github.com/step-security/dev-machine-guard/actions/workflows/gosec.yml/badge.svg?branch=main" alt="Gosec"></a>
- <a href="https://github.com/step-security/dev-machine-guard/actions/workflows/release.yml"><img src="https://github.com/step-security/dev-machine-guard/actions/workflows/release.yml/badge.svg" alt="Release"></a>
- <a href="https://goreportcard.com/report/github.com/step-security/dev-machine-guard"><img src="https://goreportcard.com/badge/github.com/step-security/dev-machine-guard" alt="Go Report Card"></a>
- <a href="https://pkg.go.dev/github.com/step-security/dev-machine-guard"><img src="https://pkg.go.dev/badge/github.com/step-security/dev-machine-guard.svg" alt="Go Reference"></a>
- <a href="LICENSE"><img src="https://img.shields.io/badge/license-Apache%202.0-blue.svg" alt="License: Apache 2.0"></a>
- <a href="https://github.com/step-security/dev-machine-guard/releases"><img src="https://img.shields.io/github/v/release/step-security/dev-machine-guard?label=release&color=purple" alt="Latest release"></a>
- </p>
- <p align="center">
- <b>Scan your dev machine for AI agents, MCP servers, IDE extensions, suspicious files, and risky package-manager configs — in seconds.</b>
- </p>
- ## Why Dev Machine Guard?
- Developer machines are the new attack surface. They hold high-value assets — GitHub tokens, cloud credentials, SSH keys — and routinely execute untrusted code through dependencies and AI-powered tools. Recent supply chain attacks have shown that malicious VS Code extensions can steal credentials, rogue MCP servers can access your codebase, and compromised npm packages can exfiltrate secrets.
- <p align="center">
- <img src="images/attack-surface.png" alt="Developer machine attack surface — malicious extensions, rogue MCP servers, unvetted AI agents, compromised packages" width="800">
- </p>
- **EDR and traditional MDM solutions** monitor device posture and compliance, but they have **zero visibility** into the developer tooling layer:
- | Capability | EDR / MDM | Dev Machine Guard |
- | --------------------------- | :-------: | :---------------: |
- | IDE extension audit | | Yes |
- | AI agent & tool inventory | | Yes |
- | MCP server config audit | | Yes |
- | Package scanning (Node.js, Homebrew, Python, system) | | Yes |
- | Package-manager config audit (registry, cooldown, auth) | | Yes |
- | Suspicious file / IOC detection | | Yes |
- | Cross-platform (macOS, Windows, Linux) | Yes | Yes |
- | Device posture & compliance | Yes | |
- | Malware / virus detection | Yes | |
- **Dev Machine Guard is complementary to EDR/MDM — not a replacement.** Deploy it alongside your existing tools via MDM (Jamf, Kandji, Intune) or run it standalone.
- <p align="center">
- <img src="images/blind-spots.png" alt="The developer tooling blind spot — EDR and MDM miss IDE extensions, AI agents, MCP servers, and npm packages" width="800">
- </p>
- ## Quick Start
- The steps below install the binary directly and are intended for **community users** evaluating Dev Machine Guard on a single machine.
- **Enterprise customers should not install the binary manually.** Deploy Dev Machine Guard across your fleet using the loader script through your MDM or EDR tooling. See the [Installation Script documentation](https://docs.stepsecurity.io/developer-machines/installation-script) for the supported, auto-updating deployment flow.
- ### Install from release (community)
- Download the latest binary for your platform from [GitHub Releases](https://github.com/step-security/dev-machine-guard/releases). Release asset filenames embed the version (for example, `stepsecurity-dev-machine-guard-1.19.1-darwin`), so the snippets below resolve the latest tag first rather than hardcoding a version.
- **macOS** (universal binary — Apple Silicon and Intel)
- ```bash
- VERSION=$(curl -fsSL https://api.github.com/repos/step-security/dev-machine-guard/releases/latest | grep '"tag_name"' | cut -d'"' -f4 | sed 's/^v//')
- curl -fsSL "https://github.com/step-security/dev-machine-guard/releases/download/v${VERSION}/stepsecurity-dev-machine-guard-${VERSION}-darwin" -o stepsecurity-dev-machine-guard
- chmod +x stepsecurity-dev-machine-guard
- # Run the scan
- ./stepsecurity-dev-machine-guard
- ```
- **Windows** (PowerShell)
- ```powershell
- $version = (Invoke-RestMethod https://api.github.com/repos/step-security/dev-machine-guard/releases/latest).tag_name.TrimStart('v')
- $arch = if ($env:PROCESSOR_ARCHITECTURE -eq 'ARM64') { 'arm64' } else { 'amd64' }
- Invoke-WebRequest -Uri "https://github.com/step-security/dev-machine-guard/releases/download/v$version/stepsecurity-dev-machine-guard-$version-windows_$arch.exe" -OutFile "stepsecurity-dev-machine-guard.exe"
- # Run the scan
- .\stepsecurity-dev-machine-guard.exe
- ```
- **Linux**
- ```bash
- VERSION=$(curl -fsSL https://api.github.com/repos/step-security/dev-machine-guard/releases/latest | grep '"tag_name"' | cut -d'"' -f4 | sed 's/^v//')
- ARCH=$(uname -m | sed 's/x86_64/amd64/;s/aarch64/arm64/')
- curl -fsSL "https://github.com/step-security/dev-machine-guard/releases/download/v${VERSION}/stepsecurity-dev-machine-guard-${VERSION}-linux_${ARCH}" -o stepsecurity-dev-machine-guard
- chmod +x stepsecurity-dev-machine-guard
- # Run the scan
- ./stepsecurity-dev-machine-guard
- ```
- Pre-built `.deb` and `.rpm` packages are also available on the [releases page](https://github.com/step-security/dev-machine-guard/releases).
- ### Build from source
- ```bash
- git clone https://github.com/step-security/dev-machine-guard.git
- cd dev-machine-guard
- make build
- ./stepsecurity-dev-machine-guard
- ```
- Requires Go 1.26+. The binary has zero external dependencies.
- ## Usage
- ```
- stepsecurity-dev-machine-guard [COMMAND] [OPTIONS]
- ```
- ### Commands
- | Command | Description |
- | ---------------- | --------------------------------------------------------------- |
- | _(none)_ | Run a scan (community mode, pretty output) |
- | `configure` | Interactively set all settings (enterprise, scan, output) |
- | `configure show` | Show current configuration (API key masked) |
- | `install` | Install scheduled scanning — launchd (macOS), systemd (Linux), schtasks (Windows) |
- | `uninstall` | Remove scheduled scanning configuration |
- | `send-telemetry` | Upload scan results to the StepSecurity dashboard (enterprise) |
- ### Output Formats
- | Flag | Description |
- | ------------- | ---------------------------------------- |
- | `--pretty` | Pretty terminal output (default) |
- | `--json` | JSON output to stdout |
- | `--html FILE` | Self-contained HTML report saved to FILE |
- ### Options
- | Flag | Description |
- | ---------------------------- | ------------------------------------------------------------- |
- | `--search-dirs DIR [DIR...]` | Search DIRs instead of `$HOME` (replaces default; repeatable) |
- | `--enable-npm-scan` | Enable Node.js package scanning |
- | `--disable-npm-scan` | Disable Node.js package scanning |
- | `--enable-brew-scan` | Enable Homebrew package scanning |
- | `--disable-brew-scan` | Disable Homebrew package scanning |
- | `--enable-python-scan` | Enable Python package scanning |
- | `--disable-python-scan` | Disable Python package scanning |
- | `--include-bundled-plugins` | Include bundled/platform IDE plugins in output |
- | `--log-level=LEVEL` | Log level: `error` \| `warn` \| `info` \| `debug` |
- | `--verbose` | Shortcut for `--log-level=debug` |
- | `--force-scan` | Scan now regardless of the dashboard cadence; still checks in |
- | `--color=WHEN` | Color mode: `auto` \| `always` \| `never` (default: `auto`) |
- | `-v`, `--version` | Show version |
- | `-h`, `--help` | Show help |
- ### Examples
- ```bash
- # Pretty terminal output (default)
- ./stepsecurity-dev-machine-guard
- # JSON output
- ./stepsecurity-dev-machine-guard --json
- ./stepsecurity-dev-machine-guard --json | python3 -m json.tool # formatted
- ./stepsecurity-dev-machine-guard --json > scan.json # to file
- # HTML report
- ./stepsecurity-dev-machine-guard --html report.html
- # Verbose scan with npm packages — shows progress spinners and timing
- ./stepsecurity-dev-machine-guard --verbose --enable-npm-scan
- # Scan specific directories instead of $HOME
- ./stepsecurity-dev-machine-guard --search-dirs /Volumes/code
- ./stepsecurity-dev-machine-guard --search-dirs /tmp /opt # multiple dirs
- # Pipe JSON through jq to extract just AI tools
- ./stepsecurity-dev-machine-guard --json | jq '.ai_agents_and_tools'
- # Count IDE extensions
- ./stepsecurity-dev-machine-guard --json | jq '.summary.ide_extensions_count'
- # Check for MCP configs (exit 1 if any found — useful in CI)
- count=$(./stepsecurity-dev-machine-guard --json | jq '.summary.mcp_configs_count')
- [ "$count" -gt 0 ] && echo "MCP servers detected!" && exit 1
- # Disable colors for piping or logging
- ./stepsecurity-dev-machine-guard --color=never 2>&1 | tee scan.log
- # Enterprise: configure all settings interactively
- ./stepsecurity-dev-machine-guard configure
- # Enterprise: view saved configuration (API key masked)
- ./stepsecurity-dev-machine-guard configure show
- # Enterprise: install scheduled scanning (launchd / systemd / schtasks)
- ./stepsecurity-dev-machine-guard install
- # Enterprise: one-time telemetry upload
- ./stepsecurity-dev-machine-guard send-telemetry
- # Enterprise: scan even when the dashboard-managed scan cadence says "not due"
- # (equivalent env var: STEPSEC_FORCE_SCAN=1; disable gating: STEPSEC_DISABLE_RUN_GATE=1)
- ./stepsecurity-dev-machine-guard send-telemetry --force-scan
- # Enterprise: remove scheduled scanning
- ./stepsecurity-dev-machine-guard uninstall
- ```
- ## Configuration
- Run `configure` to set up enterprise credentials and default search directories:
- ```bash
- ./stepsecurity-dev-machine-guard configure
- ```
- This interactively prompts for all configurable settings:
- | Setting | Description | Default |
- | ------------------ | ------------------------------------------- | --------------- |
- | Device ID | Optional custom identifier for this device | Machine serial |
- | Customer ID | Your StepSecurity customer identifier | _(not set)_ |
- | API Endpoint | StepSecurity backend URL | _(not set)_ |
- | API Key | Authentication key for telemetry uploads | _(not set)_ |
- | Scan Frequency | How often scheduled scans run (hours) | _(not set)_ |
- | Search Directories | Comma-separated list of directories to scan | `$HOME` |
- | Enable NPM Scan | Node.js package scanning | `auto` |
- | Enable Brew Scan | Homebrew package scanning | `auto` |
- | Enable Python Scan | Python package scanning | `auto` |
- | Color Mode | Terminal color output | `auto` |
- | Output Format | Default output format | `pretty` |
- | HTML Output File | Default path for HTML reports | _(not set)_ |
- | Log Level | Logging verbosity | `error` |
- On an ephemeral device with no detected user identity (empty, `unknown`, or `root`), assign one once:
- ```bash
- ./stepsecurity-dev-machine-guard configure --non-interactive --user-id employee-123
- ```
- This does not require enterprise credentials. The ID uses 1–256 ASCII letters, digits, or `@ . _ + -`; surrounding whitespace is trimmed, control characters are rejected, and `unknown`/`root` are reserved regardless of case. An ID-only JSON import such as `{"user_id":"employee-123"}` also works through `configure --from-file PATH`.
- An existing detected email or regular OS username prevents assignment. Once assigned, scans and hooks use the stored ID even if automatic detection changes. Repeating the same value succeeds; replacing or clearing it fails without changing configuration. Other configure operations preserve it. The guarantee applies to the existing configuration scope; the config file owner can still edit or remove files directly. Provision after cloning an ephemeral image to avoid copying an assigned identity to every clone.
- When no user identity is available, scanning continues as `unknown` with a warning and remains eligible for later assignment. Enterprise telemetry includes explicit `user_id_missing` and `device_id_missing` booleans based on the resolved identities, including `false` values. OS login state (`no_user_logged_in`) is detected independently of attribution. Backend handling of missing device IDs requires deployment verification.
- View current settings:
- ```bash
- ./stepsecurity-dev-machine-guard configure show
- ```
- ```
- Configuration (~/.stepsecurity/config.json):
- Customer ID: my-company
- API Endpoint: https://api.stepsecurity.io
- API Key: ***a1b2
- Scan Frequency: 4 hours
- Search Directories: $HOME, /Volumes/code
- Enable NPM Scan: auto
- Enable Brew Scan: auto
- Enable Python Scan: auto
- Color Mode: auto
- Output Format: pretty
- Log Level: error
- ```
- Configuration is saved to `~/.stepsecurity/config.json` with `0600` permissions (owner read/write only).
- To label a device for the dashboard, add `"device_tag": "avd-pool-eu"` to the
- config JSON, pass `configure --device-tag avd-pool-eu`, or use the MSI
- `DEVICETAG` property. The tag is set in the agent configuration only (the
- dashboard shows it and never edits it). In the dashboard's Device activity
- settings an admin can limit retirement (on uninstall, by inactivity and from
- the dashboard) to devices carrying chosen tags; the default applies it to all
- devices.
- To use a custom device ID, add `"device_id": "my-device-123"` to the existing
- config JSON, set **Device ID** through `configure`, or pass
- `configure --device-id my-device-123`. The override is also
- supported in JSON supplied to `configure --from-file`. A literal device ID
- must follow the ID format above; empty or whitespace-only inputs are rejected.
- An absent saved override uses the machine serial as before. To reset an
- existing override, enter `-` at the interactive Device ID prompt or pass
- `--device-id -`. The agent continues
- reporting the actual machine serial in `serial_number`. WSL guest scans derive
- separate IDs from the host ID and distro ID.
- Choose an ID unique within your customer and keep it stable: changing the ID
- registers a separate device in the backend; it does not migrate existing history.
- The value may also be a template over `{serial}` and `{hostname}` (for
- example `{hostname}` on pooled virtual desktops rebuilt under stable names, so
- a rebuilt host keeps its device and history), and the `STEPSEC_DEVICE_ID`
- environment variable overrides it at run time. A device is always the machine:
- on a shared Windows host (a multi-session edition or the RD Session Host role;
- with `multi_user_detection: users`, also one where two developers have each
- scanned on two different days) every signed-in session runs its own scan and
- reports under the same device with its own `user_identity` and `user_token`,
- and the dashboard lists the users inside the device; `multi_user_detection:
- off` keeps one scan per tick. `identity --json` prints the identity a machine
- reports under. Every scan also reports whether the machine is virtual or
- physical and on which hypervisor or cloud. See
- [docs/deploying-on-virtual-desktops.md](docs/deploying-on-virtual-desktops.md).
- Every setting can also be written without prompts, which is how MDM, MSI and
- CI deployments configure the agent. Any setting flag passed to `configure`
- selects the no-prompt path; settings you don't pass keep their current value.
- Credentials are required only when the command sets one of them, passes
- `--from-file`, or sets nothing at all, so a community install can change
- preferences (`configure --color=never`) without them:
- ```bash
- ./stepsecurity-dev-machine-guard configure --non-interactive \
- --customer-id my-company --api-endpoint https://api.stepsecurity.io \
- --device-id my-device-123 --scan-frequency 4 --search-dirs /Volumes/code \
- --disable-brew-scan --log-level=info
- ```
- | Setting | `configure` flag |
- | ----------------------- | ---------------------------------------------------- |
- | User ID (one-time) | `--user-id ID` (only when no usable identity is detected) |
- | Device ID | `--device-id VALUE` (a literal, or a template over `{serial}` `{hostname}`; `-` clears) |
- | Device tag | `--device-tag TAG` (`-` clears) |
- | Retire on uninstall | `--retire-on-uninstall[=BOOL]` |
- | Multi-user detection | `--multi-user-detection MODE` (`edition`, `users`, `off`; `-` clears) |
- | Customer ID | `--customer-id ID` |
- | API Endpoint | `--api-endpoint URL` |
- | API Key | `--api-key KEY` (or `DMG_API_KEY` env var) |
- | Scan Frequency | `--scan-frequency HOURS` |
- | Max Execution Duration | `--max-execution-duration 2h` (`off` to disable) |
- | Search Directories | `--search-dirs DIR...` |
- | NPM / Brew / Python / runtime scan | `--enable-<x>-scan` / `--disable-<x>-scan` |
- | Node / Python scanner | `--legacy-node-scan` / `--disk-node-scan`, `--legacy-python-scan` / `--disk-python-scan` |
- | macOS TCC-protected dirs | `--include-tcc-protected` / `--no-include-tcc-protected` |
- | macOS network volumes | `--include-network-volumes` / `--no-include-network-volumes` |
- | Color Mode | `--color=auto\|always\|never` |
- | Output Format / HTML file | `--pretty`, `--json`, `--html FILE` |
- | Log Level | `--log-level=LEVEL` |
- | Install Directory | `--install-dir DIR` |
- | Self-update | `--auto-update=BOOL`, `--update-mirror-url URL`, `--update-lag-behind N`, `--update-cooldown-hours N` |
- | Everything at once | `--from-file config.json` (flags override the file) |
- **CLI flags always override config file values** — this matches the shell script behavior. For example, if your config has `output_format: json`, running `./stepsecurity-dev-machine-guard --pretty` uses pretty output. To clear a value during configuration, enter a single dash (`-`).
- ### Logging and Verbose Mode
- By default in community mode, progress messages (spinners, step details) are **suppressed** — you only see the final output. This keeps stdout clean for piping.
- ```bash
- # Default: quiet — clean output, no progress spinners
- ./stepsecurity-dev-machine-guard --json > scan.json
- # Verbose: show progress spinners and step timing
- ./stepsecurity-dev-machine-guard --verbose
- # Fine-grained control: set log level
- ./stepsecurity-dev-machine-guard --log-level=info
- # Save log level in config so it persists across runs
- ./stepsecurity-dev-machine-guard configure
- ```
- In enterprise mode (`send-telemetry`, `install`), progress is **always shown** regardless of the log level — the output is captured as execution logs and sent to the backend for debugging.
- ## What It Detects
- See [SCAN_COVERAGE.md](SCAN_COVERAGE.md) for the full catalog of supported detections.
- | Category | Examples |
- | -------------------- | ---------------------------------------------------------------------------------------- |
- | IDEs & Desktop Apps | VS Code, Cursor, Windsurf, Antigravity, Kiro, Zed, Claude, Copilot, JetBrains suite (13 IDEs), Eclipse, Android Studio |
- | AI CLI Tools | Claude Code, Codex, Gemini CLI, Kiro CLI, GitHub Copilot CLI, Aider, OpenCode, Cursor Agent, Pi, Factory Droid, Amp, Grok Build, Kimi Code, Muse Code, Hermes Agent, Oh My Pi |
- | AI Agents | Claude Cowork, OpenClaw, ClawdBot, GPT-Engineer |
- | AI Frameworks | Ollama, LM Studio, LocalAI, Text Generation WebUI |
- | MCP Server Configs | Claude Desktop, Claude Code, Cursor, Windsurf, Antigravity, Zed, Open Interpreter, Codex, OpenCode, GitHub Copilot, Pi |
- | Agent Plugins & Skills | Claude Code, Codex and GitHub Copilot (CLI, standalone app and VS Code) plugin installations, Pi packages and extensions, declared components, standalone Claude commands, and recorded skill-use counters |
- | IDE Extensions | VS Code, Cursor, Windsurf, Antigravity, JetBrains, Eclipse, Xcode, Android Studio |
- | Browser Extensions | Google Chrome, Microsoft Edge, Mozilla Firefox |
- | Node.js Packages | npm, yarn, pnpm, bun (opt-in) |
- | Homebrew Packages | Formulae and casks with rich metadata (opt-in) |
- | Python Packages | pip, poetry, pipenv, uv, conda, rye (opt-in) |
- | System Packages | rpm, dpkg, pacman, apk, snap, flatpak (Linux) |
- | Go Modules | `go.mod`/`go.work` declarations, vendored modules, module cache, `GOBIN` tools, recorded checksums (enterprise) |
- | Rust Packages | `Cargo.toml` declarations, `Cargo.lock`, registry cache, Git checkouts, vendored sources, `cargo install` receipts (enterprise) |
- | PHP Packages | `composer.json` requirements, `composer.lock` snapshots and Composer 1/2 installed receipts, with separate directory presence (enterprise) |
- | .NET Packages | MSBuild project and `packages.config` references, `packages.lock.json`, `project.assets.json` restores, global package cache, .NET tools, templates and `.deps.json` (enterprise) |
- | Conda Packages | `environment.yml` declarations, `@EXPLICIT`, conda-lock and pixi locks, `conda-meta` receipts in every environment, extracted package caches and `.condarc` configuration (enterprise) |
- | Java/JVM Packages | Maven `pom.xml` and Gradle version-catalog declarations, Gradle lockfiles and verification metadata, Maven local repository and Gradle module cache files, with Maven and Gradle configuration (enterprise) |
- | Swift Packages | SwiftPM and Xcode `Package.resolved` pins, managed dependencies and binary artifacts from `workspace-state.json` (including Xcode DerivedData), and cached repositories and registry downloads in the SwiftPM cache, with SwiftPM mirror and registry configuration (enterprise) |
- | Package Configs | npm (`.npmrc`), pnpm (`.npmrc`, `config.yaml`, `pnpm-workspace.yaml`), bun (`bunfig.toml`), yarn classic & berry (`.yarnrc`/`.yarnrc.yml`), pip (`pip.conf`) — effective registry, cooldown policy, and auth surface across every scope |
- | Suspicious Files | Malicious-file IOCs from StepSecurity-maintained rules — e.g. `binding.gyp` that runs during `npm install`, and editor/AI-tool config files that auto-execute on project open |
- ## Package Configs & Suspicious Files
- Beyond inventorying *what* is installed, Dev Machine Guard inspects *how* each machine is configured to pull packages, and *whether* any files associated with known attacks are present.
- ### Package-manager config auditing
- Compromised packages most often reach a machine because that machine resolves directly from a public registry with no cooldown window against freshly published versions. Dev Machine Guard audits the package-manager configuration on each machine across every scope (project, user, and global) and resolves three things per package manager:
- - **Effective registry** — the registry packages actually resolve from, accounting for configuration precedence, so you can confirm machines route through StepSecurity Secure Registry or an internal artifact manager rather than straight to the public registry.
- - **Cooldown policy** — whether a cooldown window against newly published packages is in effect.
- - **Authentication surface** — what credentials are configured against the registry.
- Configuration is read from `.npmrc` (npm), pnpm config, `bunfig.toml` (bun), `.yarnrc` / `.yarnrc.yml` (yarn classic and berry), `pip.conf` (pip), Go's `go/env` and process environment (proxy, checksum database, private-module and auth settings, redacted on the device), and Cargo's `.cargo/config.toml` files and process environment (registries, source replacement, proxy and TLS settings, redacted on the device). In enterprise mode this rolls up into the **Package Configs** view in the dashboard, where you can spot machines that are unprotected or pointed at the wrong registry.
- Composer auditing reads selected global `config.json`, root manifests and verified process overrides. It reports observed repository, proxy, TLS and plugin settings with credentials redacted. It does not infer the repository that supplied a package or execute Composer to reconstruct runtime configuration. See [PHP Packages](SCAN_COVERAGE.md#php-packages) for scope and limits.
- NuGet auditing reads project, ancestor, user and machine `NuGet.Config` files plus verified process overrides. It reports package sources, source mapping, fallback folders, proxy, TLS and audit settings with credentials redacted, and records which configuration applied to each project and restore. It never executes `dotnet` or `nuget` or reads credential-provider output. See [.NET Packages](SCAN_COVERAGE.md#net-packages) for scope and limits.
- Ruby auditing reads RubyGems `gemrc` files and Bundler global, project and process settings for the resolved developer. It reports package sources, mirrors, install and cache paths, proxy, TLS and checksum settings with credentials redacted, and records which configuration applied to each project. It never executes `ruby`, `gem` or `bundle`, and reads `~/.gem/credentials` for existence only. See [Ruby Packages](SCAN_COVERAGE.md#ruby-packages) for scope and limits.
- Conda auditing reads `.condarc` files from the system, installation root, user and environment search path, `condarc.d` fragments, saved `conda-meta/state` variables and verified `CONDA_*` process settings for the resolved developer. It reports channels, channel aliases, custom channels, environment and package directories, proxy, TLS and safety-check settings with credentials and Anaconda tokens redacted, and records which configuration applied to each environment and project. It never executes `conda`, `mamba`, `micromamba`, `pixi` or Python, and checks `~/.netrc` and the anaconda-client token store for existence only. See [Conda Packages](SCAN_COVERAGE.md#conda-packages) for scope and limits.
- Maven auditing reads user and installation `settings.xml`, settings named by `-s`/`-gs`, POM repositories, `.mvn/maven.config`, `.mvn/jvm.config`, Maven wrapper properties and verified `MAVEN_ARGS`/`MAVEN_OPTS` options. Gradle auditing reads user, project and installation `gradle.properties`, Gradle wrapper properties and the policy part of `gradle/verification-metadata.xml`. Both report repositories, mirrors, proxies, checksum, TLS, wrapper and dependency-verification settings with credentials redacted, and record which configuration applied to each project. They never execute Maven, Gradle, a wrapper or a build script, and check `settings-security.xml` for existence only. See [Java / JVM Packages](SCAN_COVERAGE.md#java--jvm-packages) for scope and limits.
- SwiftPM auditing reads user, project and legacy `mirrors.json` and `registries.json` files and a mirror file named by `SWIFTPM_MIRROR_CONFIG`. It reports mirrors, registries, authentication types and registry signing policy with credentials redacted, and records which configuration applied to each project. It never executes Swift, Xcode or Git, never reads the Keychain or SwiftPM `security` state, and checks `~/.netrc` for existence only. See [Swift Packages](SCAN_COVERAGE.md#swift-packages) for scope and limits.
- Enterprise Device Policy can also set StepSecurity Secure Registry as the sole user-level Python index for pip and uv. It manages only the resolved developer's user configuration and shared StepSecurity `.netrc` entry, keeps pip and uv results independent, and restores owned settings on an explicit policy clear. Project files, virtual environments, system configuration, environment variables, direct URLs, and Poetry are not modified.
- ### Suspicious file detection
- Some supply chain attacks plant files that trigger code execution outside the package lifecycle scripts most tools watch — for example a malicious `binding.gyp` that runs during `npm install`, or an editor configuration file that runs when a project is opened. Dev Machine Guard ships a rules-engine scanner that flags these files as IOCs and wires the results into scan telemetry. The detector streams one file at a time, so scan memory stays bounded regardless of repository size.
- The detection rules are authored and maintained by StepSecurity, so the feature works out of the box with nothing to configure. As new attack techniques are identified, the rule set is updated centrally and your machines are evaluated against the new rules automatically. In enterprise mode, flagged files surface in the **Suspicious Files** view with a confidence level and attribution to the associated attack campaign.
- ## Output Formats
- ### Pretty Terminal Output (default)
- ```bash
- ./stepsecurity-dev-machine-guard
- ```
- <p align="center">
- <img src="images/pretty-output.png" alt="Pretty terminal output showing device info, AI agents, IDEs, MCP servers, and extensions" width="700">
- </p>
- ### JSON Output
- ```bash
- ./stepsecurity-dev-machine-guard --json
- ```
- See [examples/sample-output.json](examples/sample-output.json) for the full schema, or [Reading Scan Results](docs/reading-scan-results.md) for the schema reference. Recent scans also emit package-manager configuration inventory and `pnpm_audit` / `bun_audit` / `yarn_audit` results on the wire payload.
- ### HTML Report
- ```bash
- ./stepsecurity-dev-machine-guard --html report.html
- ```
- <p align="center">
- <img src="images/html-report.png" alt="HTML report with summary cards, device info, and detailed tables" width="700">
- </p>
- ## Community vs Enterprise
- | Feature | Community (Free) | Enterprise |
- | ----------------------------- | :--------------: | :--------: |
- | AI agent & tool inventory | Yes | Yes |
- | IDE extension scanning | Yes | Yes |
- | MCP server config audit | Yes | Yes |
- | Pretty / JSON / HTML output | Yes | Yes |
- | Package scanning (Node.js, Homebrew, Python) | Opt-in | Default on |
- | System package scanning (Linux) | Yes | Yes |
- | Package-manager config auditing | Yes | Yes |
- | Suspicious file detection | Yes | Yes |
- | Interactive configuration | Yes | Yes |
- | Centralized dashboard | | Yes |
- | Policy enforcement & alerting | | Yes |
- | Scheduled scans (launchd / systemd / schtasks) | | Yes |
- | Historical trends & reporting | | Yes |
- Enterprise mode requires a StepSecurity subscription. [Start a 14-day free trial](https://www.stepsecurity.io/start-free) by installing the StepSecurity GitHub App.
- ### Enterprise Setup
- ```bash
- # 1. Configure credentials (interactive)
- ./stepsecurity-dev-machine-guard configure
- # 2. Install scheduled scanning (launchd on macOS, systemd on Linux, schtasks on Windows)
- ./stepsecurity-dev-machine-guard install
- # 3. Or run a one-time telemetry upload
- ./stepsecurity-dev-machine-guard send-telemetry
- # 4. Uninstall scheduled scanning
- ./stepsecurity-dev-machine-guard uninstall
- ```
- **Open-source commitment:** StepSecurity enterprise customers use the exact same binary from this repository. There is no separate closed-source version — all scanning capabilities are developed and maintained here in the open. Enterprise mode adds centralized infrastructure (dashboard, policy engine, alerting) on top of the same open-source scanning engine.
- ## How It Works
- <p align="center">
- <img src="images/how-it-works.png" alt="Architecture diagram — scan sources flow through the binary to terminal, JSON, HTML, or StepSecurity dashboard outputs" width="800">
- </p>
- Dev Machine Guard is a single compiled binary that scans your developer environment. Here's what it does and — importantly — what it does **not** do:
- **What it collects:**
- - Installed IDEs, AI tools, and their versions
- - IDE extension/plugin names, publishers, and versions (VS Code, Cursor, Windsurf, Antigravity, JetBrains, Eclipse, Xcode, Android Studio)
- - Browser extension records read from the browsers' own state files (Chrome, Edge, Firefox): identity, enabled state, install source, and the permissions the browser currently honours. Browsing history, cookies, saved passwords, and page content are never opened
- - MCP server configuration (server names and commands only)
- - Node.js, Homebrew, Python, and system package listings (opt-in)
- - Package-manager configuration: effective registry, cooldown policy, and authentication surface across every scope (`.npmrc`, pnpm config, `bunfig.toml`, `.yarnrc`/`.yarnrc.yml`, `pip.conf`)
- - Suspicious files flagged by StepSecurity-maintained malicious-file rules (path and matched rule only — file contents are not collected)
- Detection uses platform-specific methods: `/Applications/` and `Info.plist` on macOS, `%LOCALAPPDATA%`/`%PROGRAMFILES%` and Windows Registry on Windows, `/opt`/`/usr/share`/`.desktop` files on Linux, and `$PATH` lookups on all platforms.
- **What it does NOT collect:**
- - Source code, file contents, or project data
- - Secrets, credentials, API keys, or tokens
- - Browsing history or personal files
- - Any data from your IDE workspaces
- **In community mode**, all data stays on your machine. Nothing is sent anywhere.
- **In enterprise mode**, scan data is sent to the StepSecurity backend for centralized visibility. The source code is fully open — you can audit exactly what is collected and transmitted.
- ## Building from Source
- ```bash
- # Build
- make build
- # Run unit tests (with race detector)
- make test
- # Run integration smoke tests
- make smoke
- # Run linter
- make lint
- # Clean build artifacts
- make clean
- ```
- ### Project Structure
- ```
- cmd/stepsecurity-dev-machine-guard/ # Binary entry point
- internal/
- ├── buildinfo/ # Version and build metadata
- ├── cli/ # Argument parser
- ├── config/ # Configuration file management and configure command
- ├── detector/ # All scanners (IDE, AI CLI, agents, frameworks, MCP, extensions, Node.js) — cross-platform
- ├── device/ # Device info (hostname, serial, OS version)
- ├── executor/ # OS abstraction interface (enables mocked unit tests)
- ├── launchd/ # macOS launchd install/uninstall
- ├── lock/ # PID-file instance locking
- ├── model/ # JSON struct types
- ├── output/ # Formatters (JSON, pretty, HTML)
- ├── progress/ # Progress spinner and logging
- ├── scan/ # Community mode orchestrator
- ├── schtasks/ # Windows Task Scheduler install/uninstall
- ├── systemd/ # Linux systemd user timer install/uninstall
- └── telemetry/ # Enterprise mode orchestrator and S3 upload
- ```
- ## How It Compares
- Dev Machine Guard is **not a replacement** for dependency scanners, vulnerability databases, or endpoint security tools. It covers a different layer — the developer tooling surface — that these tools were never designed to inspect.
- | Tool Category | What It Does Well | What It Misses |
- | ----------------------------------------- | ------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------ |
- | **`npm audit` / `yarn audit`** | Flags known CVEs in declared dependencies | Has no visibility into IDEs, AI tools, MCP servers, or IDE extensions |
- | **OWASP Dep-Check / Snyk / Socket** | Deep dependency vulnerability and supply-chain risk analysis | Does not scan the broader developer tooling layer (AI agents, IDE extensions, MCP configs) |
- | **EDR / MDM (CrowdStrike, Jamf, Intune)** | Device posture, compliance, and malware detection | Zero visibility into developer-specific tooling like IDE extensions, MCP servers, or AI agent configurations |
- Dev Machine Guard fills the gap by inventorying what is actually running in your developer environment. Deploy it alongside your existing security stack for complete coverage.
- ## Known Limitations
- - **Package scanning** (Node.js, Homebrew, Python) is opt-in in community mode and results are basic (package manager detection and package/project lists). Full dependency tree analysis is available in enterprise mode.
- - **MCP config auditing** shows which tools have MCP configs (source, vendor, and config path) but does not display config file contents in community mode. Enterprise mode sends filtered config data (server names and commands only, no secrets) to the dashboard.
- - **System package scanning** (rpm, dpkg, pacman, apk, snap, flatpak) is Linux-only.
- - **Package-manager config auditing** reports the effective registry, cooldown status, and authentication surface from configuration files; it reflects configuration as written and does not intercept package installs at runtime.
- - **Suspicious file detection** flags files that match StepSecurity-maintained rules and records the path and matched rule. It does not collect or transmit file contents. Confidence levels and attack-campaign attribution surface in the enterprise dashboard.
- ## Roadmap
- Check out the [GitHub Issues](https://github.com/step-security/dev-machine-guard/issues) for planned features and improvements. Feedback and suggestions are welcome — open an issue to start a conversation.
- ## JSON Schema
- See [examples/sample-output.json](examples/sample-output.json) for a complete sample of the JSON output, or [Reading Scan Results](docs/reading-scan-results.md) for the full schema reference.
- ## Contributing
- We welcome contributions! Whether it's adding detection for a new AI tool, improving documentation, or reporting bugs.
- See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
- **Quick contribution ideas:**
- - Add a new AI tool or IDE to the detection list
- - Improve [documentation](docs/)
- - Report bugs or request features via [issues](https://github.com/step-security/dev-machine-guard/issues)
- ## Resources
- - [Changelog](CHANGELOG.md)
- - [Scan Coverage](SCAN_COVERAGE.md) — full catalog of detections
- - [Release Process](docs/release-process.md) — how releases are signed and verified
- - [Deploying via SCCM](docs/deploying-via-sccm.md) — Windows fleet rollout via Microsoft Configuration Manager (signed MSI, no PowerShell)
- - [macOS TCC Permissions](docs/macos-tcc-permissions.md) — how the agent handles Documents/Downloads/Mail TCC dirs and network volumes (container-runtime mounts), PPPC profile for MDM-pushed Full Disk Access and network-volume pre-approval, and the `include_tcc_protected` / `include_network_volumes` config fields
- - [macOS MDM packaging](packaging/macos/README.md) — ready-made PPPC `.mobileconfig` for fleet deployment
- - [Versioning](VERSIONING.md) — why the version starts at 1.8.1
- - [Security Policy](SECURITY.md) — reporting vulnerabilities
- - [Code of Conduct](CODE_OF_CONDUCT.md)
- ## License
- This project is licensed under the [Apache License 2.0](LICENSE).
- ---
- If you find Dev Machine Guard useful, please consider giving it a star. It helps others discover the project.
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.