jcoffey-dev
d0eddd2cda
Settings › Spam Filter › Local AI (CustomComponent/LocalAi) shows whether the language-model classifier is on and which model it asks. It's off until someone turns it on. Setting it up asks "Guided or manual?": guided explains what's sent (subject and text only), that the model's word is one bounded signal and never holds up mail, and recommends running Qwen3 4B Instruct 2507 locally under llama.cpp or Ollama; it then takes the model's address and the prompt, creates the model (or reuses one of the same name) and switches the classifier on with its real id. Manual goes to the ordinary forms. Turning it off is one click and keeps the model. The page also edits inbuxa:AiLimits, sending a return to a default as null so the server keeps following it. On the schema-driven forms (ai-spam-classification spec, "INBUXA Admin"): the default prompt is prefilled when the classifier is switched to Enabled, the model form suggests local addresses, a model outside the network gets the AI-2 warning (advisory, never blocking), and the classifier form says failures never hold up mail. The server's schema gains the menu link separately, after this release.
INBUXA Admin
The administration interface for the INBUXA mail server: every server setting, first-boot setup, and recovery, in the browser.
It is schema-driven. After signing in it fetches the server's schema and builds every form, list and menu from it, so it covers every setting the server has without hardcoding any of them.
Design
- One edition. Every feature the server has is available here, with
nothing held back. See the INBUXA server's
docs/spec/. - Runs anywhere, not on the mail server. INBUXA Admin is its own
deployment, never installed onto the mail server. It's pointed at the server
either at build time (
VITE_API_BASE_URL) or at deploy time:<meta name="api-base-url" content="https://mail.example.com">inindex.html. Hosted like that, it signs in as the OAuth clientinbuxa-admin, which the server registers when it's started withINBUXA_ADMIN_URLset to INBUXA Admin's address (for development,http://localhost:5173). - INBUXA's look: the logo and ihasmail's palette.
- Two-factor setup names INBUXA as the issuer, and no longer makes authenticator apps fetch a logo from a third-party site.
Developing
npm ci
npm run dev # http://localhost:5173, against VITE_API_BASE_URL in .env.development
npm run typecheck && npx eslint src/ && npx vitest run
npm run build
Keeping up with upstream
The upstream codebase's history contains no code under a proprietary license,
so this is an ordinary git fork. upstream is a fetch-only remote:
git fetch upstream --tags
git merge v1.0.12 # the next release tag
Versions
INBUXA Admin has its own dated version (inbuxa-version.json), shown with the
upstream release it's based on: INBUXA Admin 2026.9.18 (base 1.0.11).
package.json keeps upstream's version, so upstream's bumps merge cleanly.
Source code
Every build carries its own source. The interface links to it (the user menu
and the sign-in page), and the build writes it next to the app as
source.tar.gz: the exact tree the running version was built from.
License and credits
Free software under the GNU Affero General Public License, version 3.
INBUXA Admin is forked from the upstream AGPL-3.0 web administration codebase originally developed by Stalwart Labs. Their copyright notices are kept on every file inherited from it, and INBUXA's own notice is added to the files it changes. Those files are offered upstream under the AGPL-3.0-only or a proprietary license. INBUXA uses them under the AGPL-3.0 only. INBUXA isn't affiliated with or endorsed by Stalwart Labs.