AI Agents
hindsight
Agent memory that learns — so your agents get smarter over time.
★ 47.0k starsMIT
View guide →Gini: a personal agent runtime that remembers, learns skills, and reaches you with the right control in time.
gini-agent is an open-source ai agents project by Open-Curiosity (Open-Curiosity/gini-agent on GitHub). It has 1,934 GitHub stars and 774 forks, is released under the MIT license, and was last updated on GitHub on 18 July 2026.
Decide which model provider the agent will use, from OpenAI, Anthropic, Bedrock, Azure, OpenRouter, DeepSeek, Codex OAuth, or a local server.
Install Gini with the one-line installer, which on macOS also enables autostart and opens the setup page.
Complete the setup form with your chosen provider's credentials, or run gini setup then gini start on Linux.
Chat with the agent, approve tool use when prompted, and manage runs, jobs, and memory from the web app or CLI.
One Bun process per instance owns state and work, while web, CLI, mobile, MCP, and messaging clients share one authenticated API contract.
Chat, runs, tasks, approvals, traces, audit events, jobs, memories, and skills persist across sessions and clients.
Secure credential fields, sign-in handoffs into the agent's browser, choice prompts, and confirm-before-send controls appear inline instead of prose instructions.
Local embeddings, reranking, and voice-message transcription are built in, and memory is retained, recalled, and reviewed by the runtime.
Gini improves its own skills from task outcomes through a documented reward, attribution, and human-gated review process.
Multiple isolated instances with their own ports and state, plus tunnel options (including Tailscale, ngrok, and Cloudflare) and migration from OpenClaw.
Most users use the one-line installer; source runs use Bun.
git clone https://github.com/Open-Curiosity/gini-agent.git
cd gini-agentInstall dependencies with Bun, then install and start an instance derived from the directory.
bun install
bun run gini install
bun run gini startAfter install, open the web app at 127.0.0.1:7777 and set a provider, for example OpenAI.
curl -fsSL https://raw.githubusercontent.com/Open-Curiosity/gini-agent/main/scripts/install.sh | bash
gini provider set openai gpt-5.4-mini # model names per README provider sectionIt is maintained by Open-Curiosity on GitHub.
Yes, Gini is MIT-licensed and free to run. You pay your chosen model provider; local OpenAI-compatible servers are supported if you want no API spend.
Usually yes: most providers need an API key or OAuth login (Codex), though the local provider can use a self-hosted model server with no key.
Chatbot-style agents keep state in the conversation; Gini's runtime is the system of record, with persistent tasks, approvals, memory, and audit events shared across web, CLI, and mobile clients.
Run the one-line installer, complete the /setup page with a provider, and open the web app at http://127.0.0.1:7777.
Agent memory that learns — so your agents get smarter over time.
Real-time voice runtime that keeps AI agents talking, working, and present.
AppAgent — Tencent's multimodal agent that learns smartphone apps by exploration, then operates them.
Model-agnostic marketplace of agent skills with a harness-free canonical layer.
YAO: rigorous engineering, evaluation, and governance for reusable agent skills.
Open-source AI assistant framework with skills and an OpenClaw-style agent core.
The most-starred other AI Agents projects in the LibreAI catalogue — the realistic shortlist if gini-agent is not the right fit.
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
Free and open — grab the source, star the repo, and start creating.