osquery is a high-performance, cross-platform system instrumentation framework that exposes operating system state as a relational database. It allows operators to query any aspect of a running system — processes, users, files, network sockets, hardware, events — using standard SQL, without parsing logs or writing brittle shell scripts.

Part of the Flamingo / OpenFrame ecosystem: This fork of osquery is maintained by Flamingo as a core telemetry engine inside the OpenFrame platform, providing MSPs with AI-enhanced, SQL-driven endpoint visibility across their entire client fleet.

-- Which processes are listening on port 443?
SELECT pid, name FROM processes
WHERE pid IN (SELECT pid FROM listening_ports WHERE port = 443);

-- Which users have an SSH authorized_keys entry?
SELECT username, key FROM users JOIN authorized_keys USING (uid);

• SQL interface — Full SQLite syntax over 300+ virtual OS tables
• Cross-platform — Linux, macOS, and Windows with a unified API
• Scheduled queries — Cron-like query scheduling with differential results (only log changes)
• Real-time eventing — Publisher–subscriber model for file, process, socket, and kernel events
• Distributed querying — Central orchestration of SQL across a fleet over TLS
• Extensions & IPC — Thrift-based cross-process plugin framework for custom tables and plugins
• Pluggable logging — Filesystem, TLS, Kafka, AWS Kinesis/Firehose, and custom backends
• Pluggable config — Filesystem, TLS, and extension-provided configuration sources
• Watchdog supervision — CPU/memory-aware worker supervision with automatic restart
• RocksDB persistence — Durable local state for events, query results, and metrics
• Cryptographic hashing — Streaming SHA-256, SHA-1, and MD5 for file integrity
• Secure by default — SQLite opcode allowlisting, PRAGMA restrictions, and process isolation

flowchart TD
    Config["Core Config & Flags"] --> Scheduler["Scheduler"]
    Scheduler --> SQLCore["SQL Core & Virtual Tables"]

    SQLCore --> SQLite["Embedded SQLite"]
    SQLite --> VTables["Virtual Tables (300+ OS Tables)"]
    VTables --> TablePlugins["Table Plugins / Extensions"]

    SQLCore --> DiffEngine["Diff Engine"]
    DiffEngine --> LogItem["QueryLogItem"]

    LogItem --> Logging["Logging Module"]
    LogItem --> Distributed["Distributed Querying"]

    Distributed --> RemoteHTTP["Remote HTTPS"]

    SQLCore --> Database["RocksDB Database"]

    Eventing["Eventing Core"] --> Database
    Eventing --> SQLCore

    Extensions["Extensions & IPC (Thrift)"] --> TablePlugins
    Extensions --> Config
    Extensions --> Logging

    Watcher["Init / Shutdown / Watcher"] --> Scheduler
    Watcher --> Extensions

The build process compiles a large number of bundled libraries (RocksDB, Thrift, Boost, OpenSSL, etc.) in parallel. Insufficient RAM or CPU will significantly increase build times.

Ensure you have the following installed:

• CMake 3.21+
• Git 2.x
• Python 3.6+
• Clang/LLVM 11+ (Linux/macOS) or MSVC VS 2019+ (Windows)
• Ninja 1.10+ (recommended)

Linux (Ubuntu/Debian):

sudo apt-get update && sudo apt-get install -y \
  build-essential cmake git python3 python3-pip \
  clang llvm ninja-build libstdc++-dev zlib1g-dev libssl-dev

macOS:

xcode-select --install
brew install cmake ninja python3 llvm

Windows (AMD64): Download and run the OpenFrame CLI installer:

https://github.com/flamingo-stack/openframe-cli/releases/latest/download/openframe-cli_windows_amd64.zip

# 1. Clone the repository
git clone https://github.com/flamingo-stack/osquery.git
cd osquery

# 2. Configure the build (Linux / macOS)
cmake -B build -S . -G Ninja \
  -DCMAKE_BUILD_TYPE=RelWithDebInfo

# 3. Build osqueryi (interactive shell)
cmake --build build --target osqueryi -j$(nproc)

# 4. Run the interactive shell
./build/osquery/osqueryi

Windows (PowerShell):

cmake -B build -S . -G "Visual Studio 17 2022" -A x64
cmake --build build --config RelWithDebInfo --target osqueryi
.\build\osquery\RelWithDebInfo\osqueryi.exe

-- What OS is this system running?
SELECT * FROM os_version;

-- What processes are currently running?
SELECT pid, name, path FROM processes LIMIT 10;

-- What network ports are listening?
SELECT pid, port, protocol FROM listening_ports;

-- What users exist on this system?
SELECT uid, username, shell FROM users;

-- Recent file changes (requires osqueryd with eventing)
SELECT time, action, category, path FROM file_events ORDER BY time DESC LIMIT 20;

cat > /tmp/osquery.conf << 'EOF'
{
  "options": {
    "logger_path": "/tmp/osquery_logs",
    "disable_logging": false
  },
  "schedule": {
    "os_version": {
      "query": "SELECT * FROM os_version;",
      "interval": 60
    },
    "listening_ports": {
      "query": "SELECT pid, port, protocol FROM listening_ports;",
      "interval": 30
    }
  }
}
EOF

./build/osquery/osqueryd \
  --flagfile /tmp/osquery.conf \
  --verbose

📚 See the Documentation for comprehensive guides covering:

• Getting Started — Introduction, prerequisites, quick start, and first steps
• Development Guides — Local build setup, architecture, testing, and security
• Reference Architecture — Detailed module documentation for all subsystems

All collaboration happens on the OpenMSP Slack community — not GitHub Issues or Discussions.

• 💬 Join Slack: OpenMSP Community
• 🌐 OpenMSP: https://www.openmsp.ai/
• 🦩 Flamingo: https://flamingo.run
• 🖥️ OpenFrame: https://openframe.ai

See CONTRIBUTING.md for code style, branching, commit message format, and the pull request process.