@steipete/summarize 0.9.0 → 0.10.0

package/README.md

@@ -1,17 +1,93 @@
-# Summarize 👉 Point at any URL or file. Get the gist.
+# Summarize — Chrome Side Panel + CLI
 
-Fast CLI for summarizing *anything you can point at*:
+Fast summaries from URLs, files, and media. Works in the terminal, a Chrome Side Panel and Firefox Sidebar.
 
-- Web pages (article extraction; Firecrawl fallback if sites block agents)
-- YouTube links (best-effort transcripts; can fall back to audio transcription)
-- Podcasts (Apple Podcasts / Spotify / RSS; prefers published transcripts when available; otherwise transcribes full episodes)
-- Any audio/video (local files or direct media URLs; transcribe via Whisper, then summarize)
-- Remote files (PDFs/images/audio/video via URL — downloaded and forwarded to the model)
-- Local files (PDFs/images/audio/video/text — forwarded or inlined; support depends on provider/model)
+**0.10.0 preview (unreleased):** this README reflects the upcoming release.
 
-It streams output by default on TTY and renders Markdown to ANSI (via `markdansi`) using scrollback-safe hybrid streaming (line-by-line, but buffers fenced code blocks and tables as blocks). At the end it prints a single “Finished in …” line with timing, token usage, and a best-effort cost estimate (when pricing is available).
+## 0.10.0 preview highlights (most interesting first)
 
-## Install
+- Chrome Side Panel **chat** (streaming agent + history) inside the sidebar.
+- **YouTube slides**: screenshots + OCR + transcript cards, timestamped seek, OCR/Transcript toggle.
+- Media-aware summaries: auto‑detect video/audio vs page content.
+- Streaming Markdown + metrics + cache‑aware status.
+- CLI supports URLs, files, podcasts, YouTube, audio/video, PDFs.
+
+## Feature overview
+
+- URLs, files, and media: web pages, PDFs, images, audio/video, YouTube, podcasts, RSS.
+- Slide extraction for video sources (YouTube/direct media) with OCR + timestamped cards.
+- Transcript-first media flow: published transcripts when available, Whisper fallback when not.
+- Streaming output with Markdown rendering, metrics, and cache-aware status.
+- Local, paid, and free models: OpenAI‑compatible local endpoints, paid providers, plus an OpenRouter free preset.
+- Output modes: Markdown/text, JSON diagnostics, extract-only, metrics, timing, and cost estimates.
+- Smart default: if content is shorter than the requested length, we return it as-is (use `--force-summary` to override).
+
+## Get the extension (recommended)
+
+![Summarize extension screenshot](docs/assets/summarize-extension.png)
+
+One‑click summarizer for the current tab. Chrome Side Panel + Firefox Sidebar + local daemon for streaming Markdown.
+
+**Chrome Web Store:** [Summarize Side Panel](https://chromewebstore.google.com/detail/summarize/cejgnmmhbbpdmjnfppjdfkocebngehfg)
+
+YouTube slide screenshots (from the browser):
+
+![Summarize YouTube slide screenshots](docs/assets/youtube-slides.png)
+
+### Beginner quickstart (extension)
+
+1) Install the CLI (choose one):
+   - **npm** (cross‑platform): `npm i -g @steipete/summarize`
+   - **Homebrew** (macOS arm64): `brew install steipete/tap/summarize`
+2) Install the extension (Chrome Web Store link above) and open the Side Panel.
+3) The panel shows a token + install command. Run it in Terminal:
+   - `summarize daemon install --token <TOKEN>`
+
+Why a daemon/service?
+- The extension can’t run heavy extraction inside the browser. It talks to a local background service on `127.0.0.1` for fast streaming and media tools (yt‑dlp, ffmpeg, OCR, transcription).
+- The service autostarts (launchd/systemd/Scheduled Task) so the Side Panel is always ready.
+
+If you only want the **CLI**, you can skip the daemon install entirely.
+
+Notes:
+
+- Summarization only runs when the Side Panel is open.
+- Auto mode summarizes on navigation (incl. SPAs); otherwise use the button.
+- Daemon is localhost-only and requires a shared token.
+- Autostart: macOS (launchd), Linux (systemd user), Windows (Scheduled Task).
+- Tip: configure `free` via `summarize refresh-free` (needs `OPENROUTER_API_KEY`). Add `--set-default` to set model=`free`.
+
+More:
+
+- Step-by-step install: [apps/chrome-extension/README.md](apps/chrome-extension/README.md)
+- Architecture + troubleshooting: [docs/chrome-extension.md](docs/chrome-extension.md)
+- Firefox compatibility notes: [apps/chrome-extension/docs/firefox.md](apps/chrome-extension/docs/firefox.md)
+
+### Slides (extension)
+
+- Select **Video + Slides** in the Summarize picker.
+- Slides render at the top; expand to full‑width cards with timestamps.
+- Click a slide to seek the video; toggle **Transcript/OCR** when OCR is significant.
+- Requirements: `yt-dlp` + `ffmpeg` for extraction; `tesseract` for OCR. Missing tools show an in‑panel notice.
+
+### Advanced (unpacked / dev)
+
+1) Build + load the extension (unpacked):
+   - Chrome: `pnpm -C apps/chrome-extension build`
+     - `chrome://extensions` → Developer mode → Load unpacked
+     - Pick: `apps/chrome-extension/.output/chrome-mv3`
+   - Firefox: `pnpm -C apps/chrome-extension build:firefox`
+     - `about:debugging#/runtime/this-firefox` → Load Temporary Add-on
+     - Pick: `apps/chrome-extension/.output/firefox-mv3/manifest.json`
+2) Open Side Panel/Sidebar → copy token.
+3) Install daemon in dev mode:
+   - `pnpm summarize daemon install --token <TOKEN> --dev`
+
+## CLI
+
+![Summarize CLI screenshot](docs/assets/summarize-cli.png)
+
+### Install
 
 Requires Node 22+.
 
@@ -21,7 +97,7 @@ Requires Node 22+.
 npx -y @steipete/summarize "https://example.com"
 ```
 
-- npm (global install):
+- npm (global):
 
 ```bash
 npm i -g @steipete/summarize
@@ -45,110 +121,95 @@ brew install steipete/tap/summarize
 
 Apple Silicon only (arm64).
 
-## Quickstart
+### CLI vs extension
+
+- **CLI only:** just install via npm/Homebrew and run `summarize ...` (no daemon needed).
+- **Chrome/Firefox extension:** install the CLI **and** run `summarize daemon install --token <TOKEN>` so the Side Panel can stream results and use local tools.
+
+### Quickstart
 
 ```bash
 summarize "https://example.com"
 ```
 
-## Chrome Extension (Side Panel)
-
-Want a one-click “always-on” summarizer in Chrome (real Side Panel, not injected UI)?
-
-This is a **Chrome extension** + a tiny local **daemon** (autostart service) that streams Markdown summaries for the **currently visible tab** into the Side Panel.
-
-- Step-by-step install (Chrome + daemon): `apps/chrome-extension/README.md`
-- Architecture + troubleshooting: `docs/chrome-extension.md`
-
-Quickstart (local daemon):
-
-1) Install summarize (choose one):
-   - `npm i -g @steipete/summarize`
-   - `brew install steipete/tap/summarize` (macOS arm64)
-2) Build + load the extension (unpacked):
-   - `pnpm -C apps/chrome-extension build`
-   - Chrome → `chrome://extensions` → Developer mode → “Load unpacked”
-   - Pick: `apps/chrome-extension/.output/chrome-mv3`
-3) Open the Side Panel → it shows a token + install command.
-4) Run the install command in Terminal:
-   - Installed binary: `summarize daemon install --token <TOKEN>`
-   - Repo/dev checkout: `pnpm summarize daemon install --token <TOKEN> --dev`
-5) Verify / debug:
-   - `summarize daemon status`
-   - `summarize daemon restart`
-
-Notes:
-
-- Summarization only runs when the Side Panel is open.
-- “Auto” mode summarizes on navigation (incl. SPAs); otherwise use the button.
-- The daemon is localhost-only and requires a shared token.
-- Daemon autostart: macOS (launchd), Linux (systemd user), Windows (Scheduled Task).
-- Tip: configure `free` via `summarize refresh-free` (requires `OPENROUTER_API_KEY`). Add `--set-default` to also set model=`free`, then set Model to `free` in extension settings.
-
-Troubleshooting:
-
-- **“Receiving end does not exist”**: Chrome didn’t inject the content script yet.
-  - Extension details → “Site access” → set to “On all sites” (or allow this domain)
-  - Reload the tab once.
-- **“Failed to fetch” / daemon unreachable**:
-  - Run `summarize daemon status`
-  - Check logs: `~/.summarize/logs/daemon.err.log`
+### Inputs
 
-Input can be a URL or a local file path:
+URLs or local paths:
 
 ```bash
-npx -y @steipete/summarize "/path/to/file.pdf" --model google/gemini-3-flash-preview
-npx -y @steipete/summarize "/path/to/image.jpeg" --model google/gemini-3-flash-preview
+summarize "/path/to/file.pdf" --model google/gemini-3-flash-preview
+summarize "https://example.com/report.pdf" --model google/gemini-3-flash-preview
+summarize "/path/to/audio.mp3"
+summarize "/path/to/video.mp4"
 ```
 
-Remote file URLs work the same (best-effort; the file is downloaded and passed to the model):
+YouTube (supports `youtube.com` and `youtu.be`):
 
 ```bash
-npx -y @steipete/summarize "https://example.com/report.pdf" --model google/gemini-3-flash-preview
+summarize "https://youtu.be/dQw4w9WgXcQ" --youtube auto
 ```
 
-YouTube (supports `youtube.com` and `youtu.be`):
+Podcast RSS (transcribes latest enclosure):
 
 ```bash
-npx -y @steipete/summarize "https://youtu.be/dQw4w9WgXcQ" --youtube auto
+summarize "https://feeds.npr.org/500005/podcast.xml"
 ```
 
-Podcast RSS feed (transcribes latest episode enclosure):
+Apple Podcasts episode page:
 
 ```bash
-npx -y @steipete/summarize "https://feeds.npr.org/500005/podcast.xml"
+summarize "https://podcasts.apple.com/us/podcast/2424-jelly-roll/id360084272?i=1000740717432"
 ```
 
-Apple Podcasts episode page (extracts stream URL, transcribes via Whisper):
+Spotify episode page (best-effort; may fail for exclusives):
 
 ```bash
-npx -y @steipete/summarize "https://podcasts.apple.com/us/podcast/2424-jelly-roll/id360084272?i=1000740717432"
+summarize "https://open.spotify.com/episode/5auotqWAXhhKyb9ymCuBJY"
 ```
 
-Spotify episode page (best-effort; resolves to full episode via iTunes/RSS enclosure when available — not preview clips; may fail for Spotify-exclusive shows):
+### Output length
+
+`--length` controls how much output we ask for (guideline), not a hard cap.
 
 ```bash
-npx -y @steipete/summarize "https://open.spotify.com/episode/5auotqWAXhhKyb9ymCuBJY"
+summarize "https://example.com" --length long
+summarize "https://example.com" --length 20k
 ```
 
-## What file types work?
+- Presets: `short|medium|long|xl|xxl`
+- Character targets: `1500`, `20k`, `20000`
+- Optional hard cap: `--max-output-tokens <count>` (e.g. `2000`, `2k`)
+  - Provider/model APIs still enforce their own maximum output limits.
+  - If omitted, no max token parameter is sent (provider default).
+  - Prefer `--length` unless you need a hard cap.
+- Short content: when extracted content is shorter than the requested length, the CLI returns the content as-is.
+  - Override with `--force-summary` to always run the LLM.
+- Minimums: `--length` numeric values must be >= 50 chars; `--max-output-tokens` must be >= 16.
+- Preset targets (source of truth: `packages/core/src/prompts/summary-lengths.ts`):
+  - short: target ~900 chars (range 600-1,200)
+  - medium: target ~1,800 chars (range 1,200-2,500)
+  - long: target ~4,200 chars (range 2,500-6,000)
+  - xl: target ~9,000 chars (range 6,000-14,000)
+  - xxl: target ~17,000 chars (range 14,000-22,000)
 
-This is “best effort” and depends on what your selected model/provider accepts. In practice these usually work well:
+### What file types work?
 
-- `text/*` and common structured text (`.txt`, `.md`, `.json`, `.yaml`, `.xml`, …)  
-  - text-like files are **inlined into the prompt** (instead of attached as a file part) for better provider compatibility
-- PDFs: `application/pdf` (provider support varies; Google is the most reliable in this CLI right now)
+Best effort and provider-dependent. These usually work well:
+
+- `text/*` and common structured text (`.txt`, `.md`, `.json`, `.yaml`, `.xml`, ...)
+  - Text-like files are inlined into the prompt for better provider compatibility.
+- PDFs: `application/pdf` (provider support varies; Google is the most reliable here)
 - Images: `image/jpeg`, `image/png`, `image/webp`, `image/gif`
-- Audio/Video: `audio/*`, `video/*` (when supported by the model)
+- Audio/Video: `audio/*`, `video/*` (local audio/video files MP3/WAV/M4A/OGG/FLAC/MP4/MOV/WEBM automatically transcribed, when supported by the model)
 
 Notes:
 
-- If a provider rejects a media type, the CLI fails fast with a friendly message (no “mystery stack traces”).
-- xAI models currently don’t support attaching generic files (like PDFs) via the AI SDK; use a Google/OpenAI/Anthropic model for those.
+- If a provider rejects a media type, the CLI fails fast with a friendly message.
+- xAI models do not support attaching generic files (like PDFs) via the AI SDK; use Google/OpenAI/Anthropic for those.
 
-## Model ids
+### Model ids
 
-Use “gateway-style” ids: `<provider>/<model>`.
+Use gateway-style ids: `<provider>/<model>`.
 
 Examples:
 
@@ -159,104 +220,54 @@ Examples:
 - `zai/glm-4.7`
 - `openrouter/openai/gpt-5-mini` (force OpenRouter)
 
-Note: some models/providers don’t support streaming or certain file media types. When that happens, the CLI prints a friendly error (or auto-disables streaming for that model when supported by the provider).
-
-## Output length
-
-`--length` controls *how much output we ask for* (guideline), not a hard truncation.
-
-```bash
-npx -y @steipete/summarize "https://example.com" --length long
-npx -y @steipete/summarize "https://example.com" --length 20k
-```
+Note: some models/providers do not support streaming or certain file media types. When that happens, the CLI prints a friendly error (or auto-disables streaming for that model when supported by the provider).
 
-- Presets: `short|medium|long|xl|xxl`
-- Character targets: `1500`, `20k`, `20000`
-- Optional hard cap: `--max-output-tokens <count>` (e.g. `2000`, `2k`)
-  - Provider/model APIs still enforce their own maximum output limits.
-  - If omitted, no max token parameter is sent (provider default).
-  - Prefer `--length` unless you need a hard cap (some providers count “reasoning” into the cap).
-- Minimums: `--length` numeric values must be ≥ 50 chars; `--max-output-tokens` must be ≥ 16.
-- Preset targets (source of truth: `packages/core/src/prompts/summary-lengths.ts`):
-  - short: target ~900 chars (range 600-1,200)
-  - medium: target ~1,800 chars (range 1,200-2,500)
-  - long: target ~4,200 chars (range 2,500-6,000)
-  - xl: target ~9,000 chars (range 6,000-14,000)
-  - xxl: target ~17,000 chars (range 14,000-22,000)
-
-## Limits
+### Limits
 
 - Text inputs over 10 MB are rejected before tokenization.
-- Text prompts are preflighted against the model’s input limit (LiteLLM catalog), using a GPT tokenizer.
+- Text prompts are preflighted against the model input limit (LiteLLM catalog), using a GPT tokenizer.
 
-## Common flags
+### Common flags
 
 ```bash
-npx -y @steipete/summarize <input> [flags]
+summarize <input> [flags]
 ```
 
 Use `summarize --help` or `summarize help` for the full help text.
 
 - `--model <provider/model>`: which model to use (defaults to `auto`)
 - `--model auto`: automatic model selection + fallback (default)
-- `--model <name>`: use a config-defined model (see “Configuration”)
+- `--model <name>`: use a config-defined model (see Configuration)
 - `--timeout <duration>`: `30s`, `2m`, `5000ms` (default `2m`)
 - `--retries <count>`: LLM retry attempts on timeout (default `1`)
 - `--length short|medium|long|xl|xxl|s|m|l|<chars>`
-- `--language, --lang <language>`: output language (`auto` = match source; or `en`, `de`, `english`, `german`, ...)
-- `--max-output-tokens <count>`: hard cap for LLM output tokens (optional; only sent when set)
-- `--cli [provider]`: use a CLI provider (case-insensitive; equivalent to `--model cli/<provider>`). If omitted, uses auto selection with CLI enabled.
+- `--language, --lang <language>`: output language (`auto` = match source)
+- `--max-output-tokens <count>`: hard cap for LLM output tokens
+- `--cli [provider]`: use a CLI provider (`--model cli/<provider>`). If omitted, uses auto selection with CLI enabled.
 - `--stream auto|on|off`: stream LLM output (`auto` = TTY only; disabled in `--json` mode)
-- `--plain`: Keep raw output (no ANSI/OSC Markdown rendering)
+- `--plain`: keep raw output (no ANSI/OSC Markdown rendering)
 - `--no-color`: disable ANSI colors
+- `--theme <name>`: CLI theme (`aurora`, `ember`, `moss`, `mono`)
 - `--format md|text`: website/file content format (default `text`)
-- `--markdown-mode off|auto|llm|readability`: Markdown conversion mode (default `readability`). For websites: HTML→Markdown conversion. For YouTube transcripts: `llm` formats the raw transcript into clean Markdown (headings/paragraphs).
-- `--preprocess off|auto|always`: controls `uvx markitdown` usage (default `auto`; `always` forces file preprocessing)
+- `--markdown-mode off|auto|llm|readability`: HTML -> Markdown mode (default `readability`)
+- `--preprocess off|auto|always`: controls `uvx markitdown` usage (default `auto`)
   - Install `uvx`: `brew install uv` (or https://astral.sh/uv/)
-- `--extract`: print extracted content and exit (no summary) — only for URLs
+- `--extract`: print extracted content and exit (URLs only)
   - Deprecated alias: `--extract-only`
+- `--slides`: extract slides for YouTube/direct video URLs and render them inline in the summary narrative (auto-renders inline in supported terminals)
+- `--slides-ocr`: run OCR on extracted slides (requires `tesseract`)
+- `--slides-dir <dir>`: base output dir for slide images (default `./slides`)
+- `--slides-scene-threshold <value>`: scene detection threshold (0.1-1.0)
+- `--slides-max <count>`: maximum slides to extract (default `6`)
+- `--slides-min-duration <seconds>`: minimum seconds between slides
 - `--json`: machine-readable output with diagnostics, prompt, `metrics`, and optional summary
 - `--verbose`: debug/diagnostics on stderr
-- `--metrics off|on|detailed`: metrics output (default `on`; `detailed` adds a compact 2nd-line breakdown on stderr)
+- `--metrics off|on|detailed`: metrics output (default `on`)
 
-## Verified podcast services (2025-12-25)
-
-Run: `summarize <url>`
-
-- Apple Podcasts
-- Spotify
-- Amazon Music / Audible podcast pages
-- Podbean
-- Podchaser
-- RSS feeds (Podcasting 2.0 transcripts when available)
-- Embedded YouTube podcast pages (e.g. JREPodcast)
-
-Transcription: prefers local `whisper.cpp` when installed; otherwise uses OpenAI Whisper or FAL when keys are set.
-
-## Translation paths
-
-`--language/--lang` controls the *output language* of the summary (and other LLM-generated text). Default is `auto` (match source language).
-
-When the input is audio/video, the CLI needs a transcript first. The transcript comes from one of these paths:
-
-1. **Existing transcript (preferred)**
-   - YouTube: uses `youtubei` / `captionTracks` when available.
-   - Podcasts: uses Podcasting 2.0 RSS `<podcast:transcript>` (JSON/VTT) when the feed publishes it.
-2. **Whisper transcription (fallback)**
-   - YouTube: falls back to `yt-dlp` (audio download) + Whisper transcription when configured; Apify is a last-last resort (requires `APIFY_API_TOKEN`).
-   - Prefers local `whisper.cpp` when installed + model available.
-   - Otherwise uses cloud Whisper (OpenAI `OPENAI_API_KEY`) or FAL (`FAL_KEY`) depending on configuration.
-
-For “any video/audio file” (local path or direct media URL), use `--video-mode transcript` to force “transcribe → summarize”:
-
-```bash
-summarize /path/to/file.mp4 --video-mode transcript --lang en
-```
-
-## Auto model ordering
+### Auto model ordering
 
 `--model auto` builds candidate attempts from built-in rules (or your `model.rules` overrides).
-CLI tools are **not** used in auto mode unless you explicitly enable them via `cli.enabled` in config.
+CLI tools are not used in auto mode unless you enable them via `cli.enabled` in config.
 Why: CLI adds ~4s latency per attempt and higher variance.
 Shortcut: `--cli` (with no provider) uses auto selection with CLI enabled.
 
@@ -280,28 +291,31 @@ Disable CLI attempts:
 }
 ```
 
-Note: when `cli.enabled` is set, it’s also an allowlist for explicit `--cli` / `--model cli/...`.
+Note: when `cli.enabled` is set, it is also an allowlist for explicit `--cli` / `--model cli/...`.
 
-## Website extraction (Firecrawl + Markdown)
+### Website extraction (Firecrawl + Markdown)
 
-Non-YouTube URLs go through a “fetch → extract” pipeline. When the direct fetch/extraction is blocked or too thin, `--firecrawl auto` can fall back to Firecrawl (if configured).
+Non-YouTube URLs go through a fetch -> extract pipeline. When direct fetch/extraction is blocked or too thin,
+`--firecrawl auto` can fall back to Firecrawl (if configured).
 
 - `--firecrawl off|auto|always` (default `auto`)
 - `--extract --format md|text` (default `text`; if `--format` is omitted, `--extract` defaults to `md` for non-YouTube URLs)
-- `--markdown-mode off|auto|llm|readability` (default `readability`; for non-YouTube URLs this controls HTML→Markdown conversion)
+- `--markdown-mode off|auto|llm|readability` (default `readability`)
   - `auto`: use an LLM converter when configured; may fall back to `uvx markitdown`
   - `llm`: force LLM conversion (requires a configured model key)
   - `off`: disable LLM conversion (still may return Firecrawl Markdown when configured)
 - Plain-text mode: use `--format text`.
 
-## YouTube transcripts
+### YouTube transcripts
 
-`--youtube auto` tries best-effort web transcript endpoints first. When captions aren't available, it falls back to:
+`--youtube auto` tries best-effort web transcript endpoints first. When captions are not available, it falls back to:
 
-1. **Apify** (if `APIFY_API_TOKEN` is set): Uses a scraping actor (`faVsWy9VTSNVIhWpR`)
-2. **yt-dlp + Whisper** (if `yt-dlp` is available): Downloads audio via yt-dlp, transcribes with local `whisper.cpp` when installed (preferred), otherwise falls back to OpenAI (`OPENAI_API_KEY`) or FAL (`FAL_KEY`)
+1. Apify (if `APIFY_API_TOKEN` is set): uses a scraping actor (`faVsWy9VTSNVIhWpR`)
+2. yt-dlp + Whisper (if `yt-dlp` is available): downloads audio, then transcribes with local `whisper.cpp` when installed
+   (preferred), otherwise falls back to OpenAI (`OPENAI_API_KEY`) or FAL (`FAL_KEY`)
 
 Environment variables for yt-dlp mode:
+
 - `YT_DLP_PATH` - optional path to yt-dlp binary (otherwise `yt-dlp` is resolved via `PATH`)
 - `SUMMARIZE_WHISPER_CPP_MODEL_PATH` - optional override for the local `whisper.cpp` model file
 - `SUMMARIZE_WHISPER_CPP_BINARY` - optional override for the local binary (default: `whisper-cli`)
@@ -311,17 +325,82 @@ Environment variables for yt-dlp mode:
 
 Apify costs money but tends to be more reliable when captions exist.
 
+### Slide extraction (YouTube + direct video URLs)
+
+Extract slide screenshots (scene detection via `ffmpeg`) and optional OCR:
+
+```bash
+summarize "https://www.youtube.com/watch?v=..." --slides
+summarize "https://www.youtube.com/watch?v=..." --slides --slides-ocr
+```
+
+Outputs are written under `./slides/<sourceId>/` (or `--slides-dir`). OCR results are included in JSON output
+(`--json`) and stored in `slides.json` inside the slide directory. When scene detection is too sparse, the
+extractor also samples at a fixed interval to improve coverage.
+When using `--slides`, supported terminals (kitty/iTerm/Konsole) render inline thumbnails automatically inside the
+summary narrative (the model inserts `[slide:N]` markers). Timestamp links are clickable when the terminal supports
+OSC-8 (YouTube/Vimeo/Loom/Dropbox). If inline images are unsupported, Summarize prints a note with the on-disk
+slide directory.
+
+Use `--slides --extract` to print the full timed transcript and insert slide images inline at matching timestamps.
+
 Format the extracted transcript as Markdown (headings + paragraphs) via an LLM:
 
 ```bash
 summarize "https://www.youtube.com/watch?v=..." --extract --format md --markdown-mode llm
 ```
 
-## Media transcription (Whisper)
+### Media transcription (Whisper)
+
+Local audio/video files are transcribed first, then summarized. `--video-mode transcript` forces
+direct media URLs (and embedded media) through Whisper first. Prefers local `whisper.cpp` when available; otherwise requires
+`OPENAI_API_KEY` or `FAL_KEY`.
 
-`--video-mode transcript` forces audio/video inputs (local files or direct media URLs) through Whisper first, then summarizes the transcript text. Prefers local `whisper.cpp` when available; otherwise requires `OPENAI_API_KEY` or `FAL_KEY`.
+### Local ONNX transcription (Parakeet/Canary)
 
-## Configuration
+Summarize can use NVIDIA Parakeet/Canary ONNX models via a local CLI you provide. Auto selection (default) prefers ONNX when configured.
+
+- Setup helper: `summarize transcriber setup`
+- Install `sherpa-onnx` from upstream binaries/build (Homebrew may not have a formula)
+- Auto selection: set `SUMMARIZE_ONNX_PARAKEET_CMD` or `SUMMARIZE_ONNX_CANARY_CMD` (no flag needed)
+- Force a model: `--transcriber parakeet|canary|whisper|auto`
+- Docs: `docs/nvidia-onnx-transcription.md`
+
+### Verified podcast services (2025-12-25)
+
+Run: `summarize <url>`
+
+- Apple Podcasts
+- Spotify
+- Amazon Music / Audible podcast pages
+- Podbean
+- Podchaser
+- RSS feeds (Podcasting 2.0 transcripts when available)
+- Embedded YouTube podcast pages (e.g. JREPodcast)
+
+Transcription: prefers local `whisper.cpp` when installed; otherwise uses OpenAI Whisper or FAL when keys are set.
+
+### Translation paths
+
+`--language/--lang` controls the output language of the summary (and other LLM-generated text). Default is `auto`.
+
+When the input is audio/video, the CLI needs a transcript first. The transcript comes from one of these paths:
+
+1. Existing transcript (preferred)
+   - YouTube: uses `youtubei` / `captionTracks` when available.
+   - Podcasts: uses Podcasting 2.0 RSS `<podcast:transcript>` (JSON/VTT) when the feed publishes it.
+2. Whisper transcription (fallback)
+   - YouTube: falls back to yt-dlp (audio download) + Whisper transcription when configured; Apify is a last resort.
+   - Prefers local `whisper.cpp` when installed + model available.
+   - Otherwise uses cloud Whisper (OpenAI `OPENAI_API_KEY`) or FAL (`FAL_KEY`).
+
+For direct media URLs, use `--video-mode transcript` to force transcribe -> summarize:
+
+```bash
+summarize https://example.com/file.mp4 --video-mode transcript --lang en
+```
+
+### Configuration
 
 Single config location:
 
@@ -331,7 +410,8 @@ Supported keys today:
 
 ```json
 {
-  "model": { "id": "openai/gpt-5-mini" }
+  "model": { "id": "openai/gpt-5-mini" },
+  "ui": { "theme": "ember" }
 }
 ```
 
@@ -345,14 +425,28 @@ Shorthand (equivalent):
 
 Also supported:
 
-- `model: { "mode": "auto" }` (automatic model selection + fallback; see `docs/model-auto.md`)
+- `model: { "mode": "auto" }` (automatic model selection + fallback; see [docs/model-auto.md](docs/model-auto.md))
 - `model.rules` (customize candidates / ordering)
 - `models` (define presets selectable via `--model <preset>`)
+- `cache.media` (media download cache: TTL 7 days, 2048 MB cap by default; `--no-media-cache` disables)
 - `media.videoMode: "auto"|"transcript"|"understand"`
+- `slides.enabled` / `slides.max` / `slides.ocr` / `slides.dir` (defaults for `--slides`)
+- `ui.theme: "aurora"|"ember"|"moss"|"mono"`
 - `openai.useChatCompletions: true` (force OpenAI-compatible chat completions)
 
-Note: the config is parsed leniently (JSON5), but **comments are not allowed**.
-Unknown keys are ignored.
+Note: the config is parsed leniently (JSON5), but comments are not allowed. Unknown keys are ignored.
+
+Media cache defaults:
+
+```json
+{
+  "cache": {
+    "media": { "enabled": true, "ttlDays": 7, "maxMb": 2048, "verify": "size" }
+  }
+}
+```
+
+Note: `--no-cache` bypasses summary caching only (LLM output). Extract/transcript caches still apply. Use `--no-media-cache` to skip media files.
 
 Precedence:
 
@@ -361,7 +455,14 @@ Precedence:
 3) `~/.summarize/config.json`
 4) default (`auto`)
 
-## Environment variables
+Theme precedence:
+
+1) `--theme`
+2) `SUMMARIZE_THEME`
+3) `~/.summarize/config.json` (`ui.theme`)
+4) default (`aurora`)
+
+### Environment variables
 
 Set the key matching your chosen `--model`:
 
@@ -369,43 +470,44 @@ Set the key matching your chosen `--model`:
 - `ANTHROPIC_API_KEY` (for `anthropic/...`)
 - `XAI_API_KEY` (for `xai/...`)
 - `Z_AI_API_KEY` (for `zai/...`; supports `ZAI_API_KEY` alias)
-- `GEMINI_API_KEY` (for `google/...`)  
+- `GEMINI_API_KEY` (for `google/...`)
   - also accepts `GOOGLE_GENERATIVE_AI_API_KEY` and `GOOGLE_API_KEY` as aliases
 
 OpenAI-compatible chat completions toggle:
 
 - `OPENAI_USE_CHAT_COMPLETIONS=1` (or set `openai.useChatCompletions` in config)
 
+UI theme:
+
+- `SUMMARIZE_THEME=aurora|ember|moss|mono`
+- `SUMMARIZE_TRUECOLOR=1` (force 24-bit ANSI)
+- `SUMMARIZE_NO_TRUECOLOR=1` (disable 24-bit ANSI)
+
 OpenRouter (OpenAI-compatible):
 
 - Set `OPENROUTER_API_KEY=...`
-- Prefer forcing OpenRouter per model id: `--model openrouter/<author>/<slug>` (e.g. `openrouter/meta-llama/llama-3.1-8b-instruct:free`)
-- Built-in preset: `--model free` (uses a default set of OpenRouter `:free` models).
+- Prefer forcing OpenRouter per model id: `--model openrouter/<author>/<slug>`
+- Built-in preset: `--model free` (uses a default set of OpenRouter `:free` models)
 
 ### `summarize refresh-free`
 
 Quick start: make free the default (keep `auto` available)
 
 ```bash
-# writes ~/.summarize/config.json (models.free) and sets model="free"
 summarize refresh-free --set-default
-
-# now this defaults to free models
 summarize "https://example.com"
-
-# whenever you want best quality instead
 summarize "https://example.com" --model auto
 ```
 
-Regenerates the `free` preset (writes `models.free` into `~/.summarize/config.json`) by:
+Regenerates the `free` preset (`models.free` in `~/.summarize/config.json`) by:
 
 - Fetching OpenRouter `/models`, filtering `:free`
-- Skipping models that look very small (<27B by default) based on the model id/name (best-effort heuristic)
+- Skipping models that look very small (<27B by default) based on the model id/name
 - Testing which ones return non-empty text (concurrency 4, timeout 10s)
-- Picking a mix of “smart-ish” (bigger `context_length` / output cap) and fast models
-- Refining timings for the final selection and writing the sorted list back
+- Picking a mix of smart-ish (bigger `context_length` / output cap) and fast models
+- Refining timings and writing the sorted list back
 
-If `--model free` stops working (rate limits, allowed-provider restrictions, models removed), run:
+If `--model free` stops working, run:
 
 ```bash
 summarize refresh-free
@@ -414,7 +516,7 @@ summarize refresh-free
 Flags:
 
 - `--runs 2` (default): extra timing runs per selected model (total runs = 1 + runs)
-- `--smart 3` (default): how many “smart-first” picks (rest filled by fastest)
+- `--smart 3` (default): how many smart-first picks (rest filled by fastest)
 - `--min-params 27b` (default): ignore models with inferred size smaller than N billion parameters
 - `--max-age-days 180` (default): ignore models older than N days (set 0 to disable)
 - `--set-default`: also sets `"model": "free"` in `~/.summarize/config.json`
@@ -426,7 +528,7 @@ OPENROUTER_API_KEY=sk-or-... summarize "https://example.com" --model openrouter/
 ```
 
 If your OpenRouter account enforces an allowed-provider list, make sure at least one provider
-is allowed for the selected model. (When routing fails, `summarize` prints the exact providers to allow.)
+is allowed for the selected model. When routing fails, `summarize` prints the exact providers to allow.
 
 Legacy: `OPENAI_BASE_URL=https://openrouter.ai/api/v1` (and either `OPENAI_API_KEY` or `OPENROUTER_API_KEY`) also works.
 
@@ -442,14 +544,14 @@ Optional services:
 - `FAL_KEY` (FAL AI API key for audio transcription via Whisper)
 - `APIFY_API_TOKEN` (YouTube transcript fallback)
 
-## Model limits
+### Model limits
 
 The CLI uses the LiteLLM model catalog for model limits (like max output tokens):
 
 - Downloaded from: `https://raw.githubusercontent.com/BerriAI/litellm/main/model_prices_and_context_window.json`
 - Cached at: `~/.summarize/cache/`
 
-## Library usage (optional)
+### Library usage (optional)
 
 Recommended (minimal deps):
 
@@ -461,9 +563,30 @@ Compatibility (pulls in CLI deps):
 - `@steipete/summarize/content`
 - `@steipete/summarize/prompts`
 
-## Development
+### Development
 
 ```bash
 pnpm install
 pnpm check
 ```
+
+## More
+
+- Docs index: [docs/README.md](docs/README.md)
+- CLI providers and config: [docs/cli.md](docs/cli.md)
+- Auto model rules: [docs/model-auto.md](docs/model-auto.md)
+- Website extraction: [docs/website.md](docs/website.md)
+- YouTube handling: [docs/youtube.md](docs/youtube.md)
+- Media pipeline: [docs/media.md](docs/media.md)
+- Config schema and precedence: [docs/config.md](docs/config.md)
+
+## Troubleshooting
+
+- "Receiving end does not exist": Chrome did not inject the content script yet.
+  - Extension details -> Site access -> On all sites (or allow this domain)
+  - Reload the tab once.
+- "Failed to fetch" / daemon unreachable:
+  - `summarize daemon status`
+  - Logs: `~/.summarize/logs/daemon.err.log`
+
+License: MIT