- What does the Statamic Auto Alt Text addon actually do?
- It generates descriptive alt text for images in a Statamic site by sending them to an LLM through Prism, the PHP LLM client. The addon fills in alt fields that would otherwise be empty, so your assets stay screen-reader friendly and search engines get meaningful image context without editors writing descriptions by hand. It is designed as a convenience layer on top of Statamic's asset system, so your templates and fieldset setup stay unchanged.
- How do I install el-schneider/statamic-auto-alt-text with Composer?
- Run composer require el-schneider/statamic-auto-alt-text from your Statamic project root. Statamic addons are auto-discovered, so no manual service provider registration is normally required. After installing, publish the addon configuration with php artisan vendor:publish and add the credentials for your chosen LLM provider. Then generate alt text for a test asset to confirm the request chain reaches your provider successfully.
- Which Statamic and Laravel versions does this package support?
- The addon targets Statamic v5 and v6, so your Laravel version must match whatever those Statamic releases require. Because it depends on Prism, a pre-1.0 package, its Composer constraint tracks Prism minors directly, which means it may need a recent patch release to resolve alongside sibling packages pinning a different Prism minor. Check the composer.json of the release you install before upgrading Laravel or Statamic in the same commit.
- How do I configure the AI provider and API key for alt text generation?
- Configuration lives in the addon's published config file, where you select the Prism driver and model used for image descriptions. Provider credentials come from environment variables, so keep the API key out of version control and set it per environment. If you run multiple environments, verify the key is present in staging and production too, otherwise generation silently fails when an editor uploads an image.
- Will the addon overwrite alt text I have already written manually?
- The addon's purpose is to fill missing alt fields, so the intended workflow is that hand-written descriptions are left alone and only gaps are filled. Still, verify the behaviour on your own field configuration before enabling it across a large asset library, since a customized asset blueprint can change which field the addon writes to. For critical hero or product images, review the generated text and replace it where accuracy matters.
- Can I generate alt text for images that already exist in my asset library?
- Yes, the typical workflow is to backfill existing assets rather than only handling new uploads. For a large library this means many LLM calls, so run it in batches and monitor usage rather than firing off the whole container at once. Take a database or asset backup first so you can revert if a provider returns low-quality descriptions.
- Does generating alt text slow down image uploads or front-end requests?
- Each description is an outbound LLM request, so generation adds latency wherever it runs. Keep it out of the front-end request path entirely and prefer running it after upload or through a queued job so editors are not waiting on a provider round trip. If your queue workers are not running in an environment, generation will appear to hang or never complete.
- What does it cost to run and which LLM providers can I use?
- Cost depends entirely on the Prism driver and model you configure, since you pay your provider per image and per token. Vision-capable models cost more than text-only ones, so estimate against your expected upload volume before enabling it site-wide. Also account for retries and any bulk backfill, which can dwarf day-to-day usage.
- Is the generated alt text good enough for WCAG accessibility compliance?
- Automated alt text is a solid baseline that removes the most common failure, which is a missing alt attribute, but it is not a compliance guarantee on its own. Review output for images where the context matters, such as charts, product shots, and screenshots, and mark purely decorative images with an empty alt rather than a generated description. Treat the addon as a first pass that a human editor signs off on.
- How do I test the addon without making real API calls in my test suite?
- Mock the transport or swap in a fake Prism client through the service container so tests never hit a live provider. Assert on the resolved request and the resulting alt field value rather than on container resolution alone, which catches signature drift between Prism minors. Because the package tracks Prism minors directly, include a test that exercises the real call path so a Prism bump fails in CI instead of mid-request in production.