Back to directory

Gloamy

An early-preview, local-first personal AI agent that runs tasks on a computer through explicit subsystem contracts for model providers, messaging channels, tools, memory and device peripherals.

Gloamy

Gloamy executes tasks on a user's behalf, including chatting in a terminal, running in the background to handle incoming messages, controlling desktop applications under set permissions, running tools and scripts locally or online, remembering information between sessions, and connecting to external devices and smart hardware. It is built around explicit, swappable subsystem contracts: Provider for model backends, Channel for messaging platforms, Tool for execution surfaces, Memory for persistence, Observer for observability, RuntimeAdapter for runtime isolation and Peripheral for hardware integrations, so any of these pieces can be replaced independently.

Out of the box it supports local command-line use, channel-based operation through Telegram, Discord, Slack, Matrix, WhatsApp, iMessage, email, IRC and Nostr, both OpenAI-compatible and other model providers, and SQLite, Markdown, Lucid, PostgreSQL or no-op memory backends. A companion web application is included.

It is written in Rust and released under the MIT license; the project describes itself as an early preview with breaking changes expected before a stable release.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.