Signetai, developed by Signet AI, functions as a local-first memory and context layer specifically designed for AI agents. Categorized as an MCP (Model Context Protocol) server, it helps AI agents retain and utilize information across various interactions. Built with TypeScript and using the Bun JavaScript runtime, Signetai stores user-owned data including identity files, skills, configuration, source artifacts, session records, and a SQLite database. It aims to provide persistent knowledge and secure secrets management for diverse AI applications and users.

What does Signetai connect to?

Signetai connects to numerous AI harnesses and models, serving as a local-first memory and context layer. It integrates with AI harnesses such as Claude Code, OpenCode, OpenClaw, Codex, Hermes Agent, Pi, Oh My Pi, and Gemini CLI. Beyond agents, it connects to data sources like Obsidian vaults via a real-time file watcher, Discord through a real-time crawler, and GitHub for issues, pull requests, and discussions. This broad connectivity allows it to gather and manage diverse information for AI agents.

Core features and tools

Signetai exposes various callable tools for interaction. CLI commands include "signet setup", "signet status", "signet dashboard", "signet remember", and "signet api-key create". HTTP API endpoints are available, such as "/api/hooks/remember", "/api/memory/recall", and "/api/memory/remember". For specific integrations, it provides LLM-callable tools like signet_recall, signet_source_search, signet_session_search, and signet_remember for Pi, and "/remember" and "/recall" as native tools for OpenCode.

Who is it for?

Signetai serves a diverse audience, including Company brains seeking to connect existing tools for agents, Developers managing projects across various models and harnesses, Individuals for research, journaling, and daily work, Autonomous agents requiring continuity between unattended runs, and Agent builders needing pre-built memory infrastructure. It supports scenarios where AI agents benefit from persistent, shared knowledge. Skip this if your AI applications do not require a local-first memory or context layer.

How to get started

The server authenticates using an API key, particularly for remote daemon setups. The daemon communicates via an HTTP API, which harness integrations call.

Use cases

• Persistent, shared memory across AI agents/harnesses: Enables different AI tools to access a consistent knowledge base, system prompts, and session transcripts across multiple sessions.
• Automated knowledge graph creation and recall: Distills raw content and imported files into structured memory, forming a living semantic ontology through a background process called "dreaming."
• Secure secrets management: Provides a mechanism for AI agents to securely access credentials and sensitive information, utilizing encrypted local storage and 1Password integration.

Signetai offers a foundational memory and context layer for AI agents, integrating with numerous harnesses and data sources. It provides essential capabilities like persistent memory, automated knowledge graph creation, and secure secrets management. Its design supports various users, from individual researchers to agent builders, ensuring continuity and secure operation for AI applications.