Home Assistant add-on install repository: this is what you point Home
Assistant's Add-on Store at. It carries only what Supervisor needs to list
and install each add-on -- its config.yaml, icon, logo, and changelog --
pointing at pre-built, publicly hosted container images. It does not
contain any add-on's source code.
In Home Assistant: Settings -> Add-ons -> Add-on Store -> ⋮ -> Repositories, add:
https://github.com/a4ng3i/ha-apps
Then install any add-on listed below from the store.
Safe, bidirectional Git sync for /config -- staged pulls,
validate-before-apply, automatic pre-pull backups, and human-in-the-loop
conflict resolution. See ha-git-sync/DOCS.md for
setup and usage.
Same add-on, built automatically from the source repo's uat (staging)
branch -- lets you install and run an upcoming release for real, side by
side with the production add-on, before it's promoted to main. Fully
independent install: its own slug, its own /data, its own SSH deploy
key. Do not point both the production and beta add-ons at the same
GitHub repo with automatic sync enabled on both -- two independent sync
engines racing to push/pull the same /config is exactly the kind of
conflict this add-on exists to prevent. Use a separate test repo for the
beta instance, or leave its Sync policy on manual (the default).
Exposes an OpenAI-compatible API backed by your own logged-in browser
sessions on consumer AI chat products (Claude, ChatGPT, Gemini, DeepSeek,
Grok, Perplexity) -- any app that speaks the OpenAI API can use your
existing chat subscriptions instead of a metered API key. Read
chatrelay/DOCS.md before
installing -- this automates consumer chat accounts, which is
typically against those products' Terms of Service even on an account
you already pay for; built for private personal use only.
Same add-on, built from the source repo's uat branch. Independent
install from production: its own slug, its own /data. Treat it as a
preview, not something to depend on.
Read-only health, security and insights analysis for Home Assistant:
reads logs, configuration, events and the recorder database, finds
bugs, misconfigurations and security gaps, and uses AI (Claude, or a
local OpenAI-compatible model) to explain fixes and what happens in your
house. Everything sent to the AI is redacted first and logged. See
ha-ai-assistant-beta/DOCS.md.
Same add-on, built from the source repo's uat branch. Independent
install from production: its own slug, its own /data. Treat it as a
preview, not something to depend on.
Each add-on's config.yaml here points at a pre-built image in GHCR
(ghcr.io/a4ng3i/...), published publicly. Installing/updating pulls
that image directly -- nothing is built on your Home Assistant
instance.