Independent AI engineering · Public by design

    Local AI,built in the open.

    ForgeGuard AI builds and maintains open-source servers and tools for running AI on hardware you control — each published with its documentation, releases, and limitations.

    Explore the projects See a running system
    4 active projectsOpen source

    What the lab builds

    Systems for running AI on your own hardware.

    An independent engineering lab focused on local AI, model-serving infrastructure, agent systems, practical tooling, and responsible open-source stewardship.

    Domain / AI runtimes

    Speech and inference

    Self-hosted, OpenAI-compatible, observable services with explicit hardware targets.

    Inspect domain
    Domain / Systems

    Agents and infrastructure

    Deployment, orchestration, privacy, observability, and controlled execution.

    Inspect domain
    Domain / Stewardship

    Open work with lineage

    Attribution, divergence, licenses, lifecycle, and limitations made visible.

    Inspect domain

    Projects

    The public catalog.

    Every project links its source, documentation, releases, and limitations. Theproject index adds filtering by category, platform, and kind.

    Kokoro Server

    Active

    OpenAI-compatible local speech synthesis for container and Kubernetes deployments.

    Category
    Speech & audio AI
    Runs on
    docker · kubernetes · cuda · jetson

    Hermes Agent

    Active

    A ForgeGuard-maintained distribution of the Hermes self-improving AI agent, as versioned container images and desktop installers.

    Category
    Agent systems
    Runs on
    docker · cli · desktop

    Markdown Live Preview

    Active

    A privacy-conscious, browser-only Markdown editor with live preview, packaged as a self-hostable container and Helm chart.

    Category
    Developer tools
    Runs on
    docker · compose · kubernetes

    Evidence

    Inspect a running system.

    Project UI and operational facts appear before broad claims about capability.

    ForgeGuard Kokoro Server web console showing voice selection and synthesis controls

    Kokoro Server

    Deployable local speech synthesis.

    OpenAI-compatible TTS with container and Helm distribution, warmup-aware health, optional bearer authentication, HTTPS support, and GPU telemetry.

    Model
    Kokoro-82M
    Release
    v1.1.0
    License
    Apache-2.0
    Status
    Active
    Open project dossier
    Also running: Faster Whisper Server — local speech recognition with the same operational contract.

    Architecture

    The operational contract.

    Clients speak a standard API to an interface layer; inference runs on local compute with health and telemetry surfaced alongside.

    Where to start

    Paths for different readers.

    01
    Hiring and leadership

    Engineering profile, operating principles, scope, and standards.

    Review     02
    Clients and collaborators

    Capabilities, engagement fit, expectations, and contact paths.

    Discuss     03
    Engineers and contributors

    Source, documentation, releases, attribution, and contribution paths.

    Explore