https://github.com/ooloth/hub
A personal dashboard combining a TUI + Claude Code session launcher that helps me observe and improve all my software and deployments.
https://github.com/ooloth/hub
Last synced: 9 days ago
JSON representation
A personal dashboard combining a TUI + Claude Code session launcher that helps me observe and improve all my software and deployments.
- Host: GitHub
- URL: https://github.com/ooloth/hub
- Owner: ooloth
- Created: 2026-04-16T04:26:59.000Z (about 2 months ago)
- Default Branch: main
- Last Pushed: 2026-05-28T02:03:55.000Z (15 days ago)
- Last Synced: 2026-05-28T03:11:43.454Z (15 days ago)
- Language: Rust
- Homepage:
- Size: 801 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 157
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- Agents: AGENTS.md
Awesome Lists containing this project
README
# Hub
A personal command center that surfaces what needs my attention today — including (especially!)
the things I didn't know were broken. Everything I'm responsible for appears in a single terminal
window, ranked by urgency so the most important signal is always at the top.
## What This Is
Other tools already surface individual signals — Sentry monitors production errors, GitHub
shows PRs, Loki aggregates logs. Hub's value is in how it combines those insights in one
prioritized view with agentic capabilities built in:
- **Urgency-ranked** — items sorted by `(urgency, age)`; the top item is always the most pressing
- **Cross-domain triage** — signals from GitHub, Linear, Loki, home servers, and any other
source appear in one ranked list; hub is the only place their urgency is compared
- **Pre-loaded investigation** — a keypress on any signal opens the right Claude Code skill
with `hub.toml` context already loaded (endpoint, query, project name); investigation
starts immediately, not after setup
- **Keypress execution** — for issues labelled `status:ready-for-agent`, a keypress in the
TUI launches an agent in a git worktree; hub handles setup so the focus is on reviewing
the result, not orchestrating the run
- **Single config source of truth** — `hub.toml` is git-ignored and per-device; onboarding
a new project to existing workflows is one file edit, no code changes; work and personal
contexts stay naturally separate
- **Launch pad, not chat** — hub surfaces signals and launches Claude Code skills or
autonomous agents preloaded with the right context; it leverages Claude Code rather than
reinventing its interface; see [Decision 007](docs/decisions/007-tui-over-web-app.md)
- **Extensible** — adding a new workflow follows a short, documented playbook spanning
`clients/`, `workflows/`, and `config/`; see [add a workflow](docs/playbooks/add-a-workflow.md)
- **Local-only** — no server, no cloud sync; each device has its own SQLite database and
runs independently
- **Rust** — two binaries: `hub` (CLI) and `hub-tui` (Ratatui dashboard); not a web app
## Docs
- [Vision](docs/vision.md) — what this is, why, and where it's going
- [Decisions](docs/decisions/) — architectural decisions and their rationale
- [Conventions](docs/conventions/) — Rust patterns used throughout
- [Playbooks](docs/playbooks/) — step-by-step guides for common tasks
- [Contributing](CONTRIBUTING.md) — setup and development instructions
- [Private Workflows](docs/architecture/private-workflows.md) — `hub-private` wiring, symlinks, and Cargo features