Presentation website and publishing

isa9000 presentation-website builds the presentation website from the podcasts folder into isa9000-presentation_website/. --serve previews it at http://127.0.0.1:8800; in the audio generation UI, Build the presentation website does the same. It needs no model credentials and never calls a model.

Listeners get search across podcasts and episodes, cover art, a player with speed control and a mini player, transcripts, download and share links, light and dark mode, a dyslexia-friendly font, and an icon for their phone’s home screen. Add "tags": ["Security", "HR"] to a podcast.json for a topic filter. The site is static files with no server code or tracking, and asks search engines not to index it.

After every change

The presentation website does not update itself. After a new podcast, a re-recorded episode, a corrected transcript, a new cover or a removed podcast folder, build it again and upload it:

isa9000 record "isa9000-audio_kb/<podcast>" --episode 2   # if an episode changed
isa9000 presentation-website
az storage blob sync --account-name <account> --container '$web' --source isa9000-presentation_website --auth-mode login --delete-destination true

Other hosts:

aws s3 sync isa9000-presentation_website s3://intranet-audio-kb --delete
gcloud storage rsync isa9000-presentation_website gs://intranet-audio-kb --recursive --delete-unmatched-destination-objects
rsync -a --delete isa9000-presentation_website/ webserver:/var/www/audio-kb/

Serve it only inside your network. The host must answer byte-range requests so the player can seek; all of the above do, Python’s http.server does not. Set [presentation_website] base_url to also get podcast feeds people can subscribe to.

Automate with GitHub Actions

Keep the podcasts folder in a private repository and let CI publish every change, from a new podcast to a single re-recorded episode. Set the repository up once:

git lfs install
git lfs track "*.mp3" "*.jpg" "*.png"
rm isa9000-audio_kb/.gitignore      # isa9000 adds it to prevent accidental commits
git add .gitattributes isa9000-audio_kb isa9000.toml   # commit the config without keys
git commit -m "Add the audio knowledge base"

.github/workflows/publish-website.yml rebuilds and uploads the presentation website whenever anything in the podcasts folder changes. The Azure identity needs the Storage Blob Data Contributor role.

name: Publish the audio knowledge base
on:
  push:
    branches: [main]
    paths: ["isa9000-audio_kb/**", "isa9000.toml"]
  workflow_dispatch:
permissions:
  contents: read
  id-token: write
jobs:
  publish:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          lfs: true
      - uses: actions/setup-python@v5
        with:
          python-version: "3.13"
      - run: pip install isa9000
      - run: isa9000 presentation-website
      - uses: azure/login@v2
        with:
          client-id: $
          tenant-id: $
          subscription-id: $
      - run: >-
          az storage blob sync --account-name $
          --container '$web' --source isa9000-presentation_website --auth-mode login --delete-destination true

.github/workflows/create-podcast.yml creates a podcast from the Actions tab and opens a pull request, so someone reviews it before it is published. Add the AZURE_OPENAI_API_KEY secret and the AZURE_OPENAI_ENDPOINT variable. On Google, commit an isa9000.toml with provider = "google" and pass a GEMINI_API_KEY secret instead.

name: Create a podcast
on:
  workflow_dispatch:
    inputs:
      request:
        description: What should people get up to speed on?
        required: true
      sources:
        description: Links, separated by spaces
        required: true
permissions:
  contents: write
  pull-requests: write
jobs:
  create:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          lfs: true
      - uses: actions/setup-python@v5
        with:
          python-version: "3.13"
      - run: pip install isa9000
      - name: Create the podcast
        env:
          AZURE_OPENAI_ENDPOINT: $
          AZURE_OPENAI_API_KEY: $
          REQUEST: $
          SOURCES: $
        run: |
          args=()
          for source in $SOURCES; do args+=(-s "$source"); done
          isa9000 create -r "$REQUEST" "${args[@]}" || [ $? -eq 3 ]
      - name: Open a pull request
        env:
          GH_TOKEN: $
          REQUEST: $
        run: |
          git config user.name "github-actions[bot]"
          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
          git switch -c "podcast/$"
          git add isa9000-audio_kb
          git commit -m "Add podcast: $REQUEST"
          git push origin HEAD
          gh pr create --title "New podcast: $REQUEST" --body "Read the transcripts and listen before merging. Without MP3s, it needs review: see podcast.json."

A podcast that needs review arrives without audio. After the review, record it on the same branch with isa9000 record and push the MP3s; merging then publishes it.