Open tools for building, running, and scheduling AI agents.
Small, self-hosted tools that came out of running a lot of CLI agents day to day. Each one solves a problem that showed up in practice rather than one that looked interesting on paper, and each runs on your own machine — there is no hosted service to sign up for and nothing phones home.
| What it does | |
|---|---|
| cronsole | One pane of glass for scheduled tasks — Windows Task Scheduler, Cronsole-native jobs, Claude Code routines and Gemini API Triggers together, with GitHub Actions and Vercel Cron as read-only observers. Local-first: the whole stack runs on your machine. → cronsole.ai-automation-tools.dev · live demo |
| agent-skills | Portable Agent Skills for Claude Code and compatible agents — each a self-contained SKILL.md an agent loads only when it needs it. → agent-skills.ai-automation-tools.dev |
| project-hub | Turns a CLI-agent workspace into a browsable local HTML console — one zero-dependency Node server over every repo, document and agent tool you own. → project-hub.ai-automation-tools.dev |
| Agent-chat | Local MCP server for AI-to-AI conversations — a SQLite-backed message bus letting two CLI agents hold a structured, turn-based conversation. → agent-chat.ai-automation-tools.dev |
| edge-spectrum | Expected returns, synthetic backtesting and odds analysis — an interactive demo plus the full self-hosted toolkit. → edge-spectrum.ai-automation-tools.dev |
| Edge-Radar | Multi-agent edge detection for prediction markets — research first, execute second. → ops page |
- Running agents on a schedule → cronsole
- Making an agent better at one task → agent-skills
- Seeing what your agents have actually produced → project-hub
- Getting two agents to talk to each other → Agent-chat
Licences differ per repo — check the LICENSE file before reusing anything; a repo without one is
all rights reserved. Security reports: see SECURITY.md.