mkzforge 2.0.0__tar.gz

mkzforge-2.0.0/.cursor/rules/project.mdc

@@ -0,0 +1,93 @@
+# Python, FFmpeg, OpenAI Whisper, YAML
+
+**You are an expert Python developer specializing in video/audio processing automation, with deep experience using FFmpeg, OpenAI Whisper, YAML-based configuration systems, and automation scripting.**
+
+## File Editing Best Practices
+
+* **Complete File Replacements**: When creating new files or when a file is in a broken state, replace the entire file content rather than attempting partial edits.
+* **Docstring and String Literals**: Always provide complete, syntactically correct blocks with explicit opening and closing quotes. Avoid partial replacements of string content.
+* **Error Recovery**: If linter errors appear after an edit, don't try to "fix" the file with additional edits. Request a complete file replacement with correct syntax.
+* **Token Efficiency**: While complete replacements use more tokens upfront, they often result in lower total cost by eliminating multiple failed attempts.
+* **Clear Context**: Always specify exact locations for changes and whether you're replacing entire files or specific sections.
+* **Syntax-Sensitive Elements**: For imports, class definitions, docstrings, and other syntax-critical code, prefer complete file replacements over partial edits.
+
+## Key Principles
+
+* Write concise, efficient Python with clear, practical examples.
+* Prioritize clarity, modularity, and automation-friendly design.
+* Favor function composition and reusable utilities over over-engineering with classes.
+* Use `subprocess` for reliable FFmpeg/CLI calls with robust error handling.
+* Follow PEP 8 and functional programming best practices.
+
+## FFmpeg Integration
+
+* Use `subprocess` or `ffmpeg-python` for invoking FFmpeg commands.
+* Validate FFmpeg command-line arguments rigorously.
+* Capture and parse FFmpeg output when needed for metadata.
+* Handle transcoding, filtering, audio processing, and stream mapping with explicit command composition.
+* Prefer raw CLI commands for flexibility when `ffmpeg-python` cannot express certain options.
+
+## OpenAI Whisper Usage
+
+* Use the OpenAI Whisper Python API or CLI wrapper effectively.
+* Automate transcription workflows with batching and error recovery.
+* Handle large audio files via chunking or segmentation if necessary.
+* Validate model selection (tiny, base, small, medium, large) based on use-case.
+* Implement automatic language detection where possible.
+* Integrate Whisper output into post-processing pipelines (e.g., subtitle generation).
+
+## YAML-Driven Workflow Design
+
+* Use `PyYAML` for safe YAML parsing and schema enforcement.
+* Design declarative YAML schemas for defining video generation pipelines.
+* Allow YAML to configure:
+
+  * Input sources
+  * FFmpeg operations (e.g., filters, trims, overlays)
+  * Whisper transcription options
+  * Output formats and destinations
+* Validate YAML structures strictly before execution.
+
+## Automation Scripting Best Practices
+
+* Build reusable utilities for command construction and execution.
+* Modularize pipeline stages (ingest, process, transcribe, output).
+* Implement dry-run modes for command preview before execution.
+* Handle exceptions explicitly and log all subprocess outputs.
+* Support CLI arguments for overrides of YAML-defined parameters.
+
+## Error Handling & Debugging
+
+* Wrap all subprocess calls with try-except blocks.
+* Decode subprocess stdout/stderr for meaningful error messages.
+* Use verbose logging during debugging and silent modes in production.
+* Implement retry logic for transient errors in long automation jobs.
+
+## Performance & Resource Management
+
+* Optimize FFmpeg pipelines with filter chains to minimize disk I/O.
+* Use temporary files or pipes when chaining processes.
+* Manage large file handling with streaming where possible.
+* Monitor system resource usage (CPU, disk I/O) during heavy processing tasks.
+
+## Dependencies
+
+* Python 3.11+
+* `PyYAML` (for YAML parsing)
+* `python-ffmpeg` (optional)
+* `subprocess` (standard)
+* OpenAI Whisper Python package
+
+## Key Conventions
+
+1. All configs and runtime parameters should be YAML-driven.
+2. Separate script concerns: configuration, command execution, post-processing.
+3. Always support verbose/debug output for troubleshooting.
+4. Document example YAML files alongside code.
+
+## Refer to
+
+* Official FFmpeg documentation: [https://ffmpeg.org/documentation.html](https://ffmpeg.org/documentation.html)
+* OpenAI Whisper GitHub: [https://github.com/openai/whisper](https://github.com/openai/whisper)
+* PyYAML documentation: [https://pyyaml.org/](https://pyyaml.org/)
+* Python `subprocess` documentation: [https://docs.python.org/3/library/subprocess.html](https://docs.python.org/3/library/subprocess.html)

mkzforge-2.0.0/.github/workflows/publish.yml

@@ -0,0 +1,32 @@
+name: Publish to PyPI
+
+on:
+  push:
+    tags:
+      - 'v*'
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Install uv and twine
+        run: pip install uv twine
+
+      - name: Build package
+        run: uv build
+
+      - name: Publish to PyPI
+        env:
+          TWINE_USERNAME: __token__
+          TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
+        run: twine upload dist/*

mkzforge-2.0.0/.gitignore

@@ -0,0 +1,59 @@
+# Compressed archives
+*.tgz
+*.tar
+*.gz
+*.bz2
+*.lz4
+
+# Media files
+*.webm
+*.mp4
+*.mp3
+*.m4a
+*.mkv
+*.mov
+*.aac
+*.ts
+*.ac3
+*.wav
+
+# Images
+*.png
+*.jpg
+*.jpeg
+*.gif
+*.xcf
+*.ttf
+*.svg
+
+# Documents
+*.docx
+*.pdf
+*.xlsx
+
+# System Files
+## Windows
+desktop.ini
+## OSX
+.DS_Store
+## IDE's
+.nbproject/
+.idea/
+
+# Python cache
+__pycache__/
+build/
+*.egg-info
+.eggs
+dist/
+lib/mkzforge/_version.py
+
+# Makefile's are generated, so no need to keep them under VCS.
+Makefile
+
+# Experiments with Natron until I'm confident it will be used.
+*.ntp
+*.ntp.~*~
+
+.venv
+uv.lock

mkzforge-2.0.0/AGENTS.md

@@ -0,0 +1,132 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Project Overview
+
+MKZ Forge is a Python video processing automation tool that wraps FFmpeg's `filter_complex` into a
+YAML-driven pipeline. It handles the full workflow: MP4→MKV normalization, silence removal,
+Whisper subtitle generation, multi-language translation, LLM metadata generation, Gemini thumbnail
+generation, and publishing.
+
+## Commands
+
+### Install & Setup
+
+```bash
+uv pip install -e . # standard editable install
+```
+
+### Running
+
+```bash
+mkzforge --help
+mkzforge new [path]           # Create new project
+mkzforge normalize            # Normalize videos (compress, silence removal, subtitles)
+mkzforge build [output]       # Build final video from mkzforge.yml
+mkzforge gensubs [video]      # Generate subtitles only
+mkzforge metadata             # Generate LLM title/description
+mkzforge genimage             # Generate Gemini thumbnail
+mkzforge publish [file]       # Publish to SFTP/YouTube/TikTok
+mkzforge serve [--workspace /path] [--http-port 9091]  # Start web interface
+```
+
+### Testing
+
+```bash
+uv run pytest                  # Run all tests, including the virtualenv.
+```
+
+Tests are split into `tests/mkzforgeunit/` (unit) and `tests/mkzforgefunc/`
+(functional/integration). Fixtures live in `tests/fixtures/configs/`.
+
+## Architecture
+
+### Pipeline Flow
+
+1. `mkzforge normalize` → converts MP4→MKV (H.265 crf=28), removes silence via FFmpeg `silencedetect`,
+   runs Whisper for subtitles, translates via Argos Translate, calls LLM for metadata
+2. `mkzforge build` → reads `mkzforge.yml`, constructs FFmpeg filter graph via `filter_complex.py`
+   DSL, produces final MP4
+3. `mkzforge publish` → uploads via SFTP (Fabric), with hooks for YouTube/TikTok
+
+### Key Modules
+
+| Module | Purpose |
+|--------|---------|
+| `cli/__init__.py` | Argument parsing, action dispatch, config merging |
+| `videos.py` | MP4→MKV, silence detection, video compilation |
+| `subtitles.py` | Whisper integration, SRT parsing/writing |
+| `i18n.py` | Argos Translate integration, subtitle re-timing |
+| `filter_complex.py` | FFmpeg filter graph DSL abstraction |
+| `metadata.py` | LangChain LLM calls for titles/descriptions |
+| `genimg.py` | Google Gemini image generation, SVG thumbnails |
+| `webserv.py` | CherryPy REST API + background workers |
+| `notify.py` | AWS SNS singleton |
+| `utils.py` | Config loading, GPU VRAM detection |
+| `types.py` | Enums: `Devices`, `Action`, `WhisperTask` |
+| `const.py` | Supported languages, API key constants |
+
+### Config Loading Order (merged, last wins)
+
+1. `/etc/mkzforge/config.yml`
+2. `~/.config/mkzforge/config.yml`
+3. `./mkzforge.yml` (project-specific)
+4. CLI arguments
+
+### Project Directory Structure
+
+```plain
+my-project/
+├── mkzforge.yml    # YAML config describing build pipeline
+├── resources/      # Input MP4/MKV files
+└── build/          # Generated outputs (MKV, SRT, MP4, PNG)
+```
+
+### `mkzforge.yml` Schema
+
+```yaml
+videos:
+  - input: [list of files/sources]
+    output: output.mp4
+    attributes: [subs, no-video, no-audio, vsync, no-publish, thumbnail]
+    languages: [en, es, fr]
+    filter_complex: [ffmpeg filter directives]
+    metadata:
+      title: "..."
+      description: "..."
+    map:
+      video: "[v]"
+      audio: "[a]"
+      subs: 0
+```
+
+Full schema documented in `lib/mkzforge/schema.json` and `doc/configuration.md`.
+
+### Whisper Model Auto-Selection (by GPU VRAM)
+
+- ≥10GB → `large-v3`; 8–10GB → `large-v2`; 6–8GB → `medium`; 3–6GB → `small`; <3GB → `base`; CPU
+  fallback → `small`
+
+### GPU Locking
+
+File-based lock prevents concurrent GPU usage across multiple `mkzforge` processes.
+
+## Key Environment Variables
+
+| Variable | Default | Purpose |
+|----------|---------|---------|
+| `LANGUAGE` | `en` | Base transcription language |
+| `LOG_LEVEL` | `INFO` | Logging verbosity |
+| `WHISPER_MODEL` | auto | Override Whisper model selection |
+| `LLM_MODEL` | `gpt-oss:20b` | LLM model name |
+| `LLM_PROVIDER` | `ollama` | LLM provider |
+| `GOOGLE_API_KEY` | — | Gemini API key |
+| `GEMINI_IMAGE_MODEL` | `gemini-2.5-flash-image` | Gemini model |
+| `HTTP_PORT` | `9091` | Web server port |
+| `OMP_NUM_THREADS` | `nproc` | CPU thread count |
+
+## Versioning
+
+Uses `setuptools-scm` — version is derived from git tags (`v*` format). Version is written to
+`lib/mkzforge/_version.py` automatically.

mkzforge-2.0.0/CLAUDE.md

@@ -0,0 +1,132 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Project Overview
+
+MKZ Forge is a Python video processing automation tool that wraps FFmpeg's `filter_complex` into a
+YAML-driven pipeline. It handles the full workflow: MP4→MKV normalization, silence removal,
+Whisper subtitle generation, multi-language translation, LLM metadata generation, Gemini thumbnail
+generation, and publishing.
+
+## Commands
+
+### Install & Setup
+
+```bash
+uv pip install -e . # standard editable install
+```
+
+### Running
+
+```bash
+mkzforge --help
+mkzforge new [path]           # Create new project
+mkzforge normalize            # Normalize videos (compress, silence removal, subtitles)
+mkzforge build [output]       # Build final video from mkzforge.yml
+mkzforge gensubs [video]      # Generate subtitles only
+mkzforge metadata             # Generate LLM title/description
+mkzforge genimage             # Generate Gemini thumbnail
+mkzforge publish [file]       # Publish to SFTP/YouTube/TikTok
+mkzforge serve [--workspace /path] [--http-port 9091]  # Start web interface
+```
+
+### Testing
+
+```bash
+uv run pytest                  # Run all tests, including the virtualenv.
+```
+
+Tests are split into `tests/mkzforgeunit/` (unit) and `tests/mkzforgefunc/`
+(functional/integration). Fixtures live in `tests/fixtures/configs/`.
+
+## Architecture
+
+### Pipeline Flow
+
+1. `mkzforge normalize` → converts MP4→MKV (H.265 crf=28), removes silence via FFmpeg `silencedetect`,
+   runs Whisper for subtitles, translates via Argos Translate, calls LLM for metadata
+2. `mkzforge build` → reads `mkzforge.yml`, constructs FFmpeg filter graph via `filter_complex.py`
+   DSL, produces final MP4
+3. `mkzforge publish` → uploads via SFTP (Fabric), with hooks for YouTube/TikTok
+
+### Key Modules
+
+| Module | Purpose |
+|--------|---------|
+| `cli/__init__.py` | Argument parsing, action dispatch, config merging |
+| `videos.py` | MP4→MKV, silence detection, video compilation |
+| `subtitles.py` | Whisper integration, SRT parsing/writing |
+| `i18n.py` | Argos Translate integration, subtitle re-timing |
+| `filter_complex.py` | FFmpeg filter graph DSL abstraction |
+| `metadata.py` | LangChain LLM calls for titles/descriptions |
+| `genimg.py` | Google Gemini image generation, SVG thumbnails |
+| `webserv.py` | CherryPy REST API + background workers |
+| `notify.py` | AWS SNS singleton |
+| `utils.py` | Config loading, GPU VRAM detection |
+| `types.py` | Enums: `Devices`, `Action`, `WhisperTask` |
+| `const.py` | Supported languages, API key constants |
+
+### Config Loading Order (merged, last wins)
+
+1. `/etc/mkzforge/config.yml`
+2. `~/.config/mkzforge/config.yml`
+3. `./mkzforge.yml` (project-specific)
+4. CLI arguments
+
+### Project Directory Structure
+
+```plain
+my-project/
+├── mkzforge.yml    # YAML config describing build pipeline
+├── resources/      # Input MP4/MKV files
+└── build/          # Generated outputs (MKV, SRT, MP4, PNG)
+```
+
+### `mkzforge.yml` Schema
+
+```yaml
+videos:
+  - input: [list of files/sources]
+    output: output.mp4
+    attributes: [subs, no-video, no-audio, vsync, no-publish, thumbnail]
+    languages: [en, es, fr]
+    filter_complex: [ffmpeg filter directives]
+    metadata:
+      title: "..."
+      description: "..."
+    map:
+      video: "[v]"
+      audio: "[a]"
+      subs: 0
+```
+
+Full schema documented in `lib/mkzforge/schema.json` and `doc/configuration.md`.
+
+### Whisper Model Auto-Selection (by GPU VRAM)
+
+- ≥10GB → `large-v3`; 8–10GB → `large-v2`; 6–8GB → `medium`; 3–6GB → `small`; <3GB → `base`; CPU
+  fallback → `small`
+
+### GPU Locking
+
+File-based lock prevents concurrent GPU usage across multiple `mkzforge` processes.
+
+## Key Environment Variables
+
+| Variable | Default | Purpose |
+|----------|---------|---------|
+| `LANGUAGE` | `en` | Base transcription language |
+| `LOG_LEVEL` | `INFO` | Logging verbosity |
+| `WHISPER_MODEL` | auto | Override Whisper model selection |
+| `LLM_MODEL` | `gpt-oss:20b` | LLM model name |
+| `LLM_PROVIDER` | `ollama` | LLM provider |
+| `GOOGLE_API_KEY` | — | Gemini API key |
+| `GEMINI_IMAGE_MODEL` | `gemini-2.5-flash-image` | Gemini model |
+| `HTTP_PORT` | `9091` | Web server port |
+| `OMP_NUM_THREADS` | `nproc` | CPU thread count |
+
+## Versioning
+
+Uses `setuptools-scm` — version is derived from git tags (`v*` format). Version is written to
+`lib/mkzforge/_version.py` automatically.

mkzforge-2.0.0/PKG-INFO

@@ -0,0 +1,245 @@
+Metadata-Version: 2.4
+Name: mkzforge
+Version: 2.0.0
+Summary: Scripts and tools to ease the processing of videos for Social Media platforms.
+Author-email: Markizano Draconus <markizano@markizano.net>
+Project-URL: Homepage, https://mkzforge.markizano.net
+Project-URL: Repository, https://github.com/markizano/mkzforge
+Project-URL: Issues, https://github.com/markizano/mkzforge/issues
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+Requires-Dist: argostranslate>=1.11.0
+Requires-Dist: boto3>=1.42.59
+Requires-Dist: cherrypy>=18.10.0
+Requires-Dist: fabric==3.2.2
+Requires-Dist: google-api-python-client>=2.193.0
+Requires-Dist: google-auth-httplib2>=0.3.0
+Requires-Dist: google-auth-oauthlib>=1.3.0
+Requires-Dist: google-genai>=1.48.0
+Requires-Dist: jsonschema>=4.25.1
+Requires-Dist: kizano==1.0.6
+Requires-Dist: langchain>=1.0.5
+Requires-Dist: langchain-ollama>=1.0.0
+Requires-Dist: numba>=0.62.1
+Requires-Dist: openai>=2.7.2
+Requires-Dist: openai-whisper==20250625
+Requires-Dist: pillow>=10.0.0
+Requires-Dist: python-ffmpeg==2.0.12
+Requires-Dist: pyyaml>=6.0.1
+Requires-Dist: requests>=2.32.5
+
+# MKZ Forge
+
+> Markizano's Forge for video production!
+
+This is a Python project I use and maintain to automate various aspects of my video editing
+process to publish on video supporting platforms like TikTok, YouTube, LinkedIn and Substack.
+
+FFMPEG is a great video editor by itself, but it's difficult to manage that incredible description
+library it has.
+
+This project handles the following process for me:
+
+* **Convert MP4 files to MKV** files to compress for disk space.
+* **Cut Silence** from videos to remove the "uhm"s and "uuh"s.
+* **Combine videos** so multiple uploads can result in a single artifact to publish.
+* **Generate subtitles** using OpenAI's `whisper` libraries for hardsubs/captions.
+* **Generate Metadata** for title and description based on the content of the video.
+* **Generate Thumbnail** using Google's GenAI Image generation API.
+* **Pristine Video Production** for that final product you see on video platforms posted by me.
+
+In a single upload (or multi-file upload), I can quickly generate videos that already have
+captions hardcoded into them, metadata attached so platforms can easily and quickly categorize
+my videos for publishing (if they support that kind of thing).
+
+With this project, I hope to make it easier for program-oriented folks like me to have a
+descriptive file that converts your videos for you instead of having to remember complex filter
+graphs. Simply describe what you want the video to do and this app will help you in executing those
+changes.
+
+This is a very context-specific application that is catered to my needs. I never expected it to
+become popular, but if it does, here's to all who may contribute to this project or find it useful
+in their own video editing process endeavours.
+
+## Usage
+
+When recording from phone or cam, it's difficult to manage all the media and content that comes in
+without paying for crazy software to get it done. With this open-source product, you can simply
+`pip install mkzforge` and go!
+
+This project honours configuration defined in a system-wide configuration file in `/etc/mkzforge/config.yml`,
+user-specific configuration defined in `~/.config/mkzforge/config.yml`, with the configurations merged
+together in that order as defined.
+
+To create a new project, let's use this:
+
+```bash
+mkzforge new
+```
+
+This will render a new project with the following directory structure:
+
+```none
+.
+├── build/
+├── readme.md
+├── resources/
+└── mkzforge.yml
+```
+
+You can drop your MP4 files from your devices into the `./resources/` directory.
+
+Next, we can run
+
+```bash
+mkzforge normalize
+```
+
+to refresh the YAML file that is the configuration driving the
+changes we will be performing here.
+
+This may take a moment as ffmpeg converts your MP4 format videos into MKV format under high
+compression as lossless as possible. This will reduce the amount of disk that is consumed by the
+videos recorded and saved raw from devices. Subtitles will also be automatically generated from
+the video files using Whisper! The system automatically detects your GPU VRAM and selects the
+best Whisper model that fits (preferring `large-v3` for high-end GPUs).
+
+**Concurrent Processing:** If you run multiple mkzforge instances simultaneously, they will
+automatically queue for GPU access using a file-based lock to prevent OOM (Out Of Memory) errors.
+Each instance waits its turn instead of competing for GPU resources.
+
+You can suppress auto-subtitle generation with the `--no-subtitles` argument.
+
+Once your videos have been compressed and subtitles generated for them, you will have artifacts
+available in the `./build/` directory as well.
+
+INFO: See doc/configuration.md for more information about `mkzforge.yml` configuration.
+
+You can use this:
+
+```bash
+mkzforge gensubs [path-to-file.mkv]
+```
+
+This will generate subtitles in `build/path-to-file.${LANG}.srt` for any video file you give this
+command. This is not necessary if you used `mkzforge refresh` to generate subs from a video
+resource. I used this to get access to the sub-generation functionality this app provided with this
+command and so I exposed that function to this command here. You can elect to do more with it after
+that.
+
+Once your videos have been compressed and your configuration updated, you will see the YAML
+has been updated with the new video. If you have a preferred name for it, you should rename
+the file prior to running `mkzforge refresh`.
+
+Observe that subtitles will also be generated as a result of this update. To avoid this, you can
+use `--no-subtitles` when executing `mkzforge refresh --no-auto-subtitles` and it will go a bit faster.
+
+You can update the YAML configuration to have it execute a number of filters and stream the videos
+together into a final cut that can be used for social media sites and such.
+
+The top-level `videos` is an array which will contain the set of video descriptions you will use
+to describe how you want transformations done on those videos.
+
+To learn more about the mkzforge.yml file, see docs/mkzforge.md
+
+Once you have your transformations written out, you can use this:
+
+```bash
+mkzforge build build/myvideo.mp4
+```
+
+This will build your video. If you omit any arguments, it'll attempt to build all videos in your
+project and ensure they are ready to go!
+
+This project takes some of the rough edges off of the `filter_complex` argument in ffmpeg.
+
+This started off as a simple script to try and automate some of the rough edges of my process
+when recording content and publishing to the platforms.
+
+## Multi-Language Subtitle Support
+
+mkzforge now supports automatic translation of subtitles to multiple languages using Argos Translate!
+
+### Quick Start
+
+1. **Configure languages in your `mkzforge.yml`:**
+
+    ```yaml
+    mkzforge:
+      language: en           # Base language for Whisper transcription
+      languages:             # Translate to these languages
+        - en                 # English (from Whisper)
+        - es                 # Spanish (translated)
+        - fr                 # French (translated)
+        - de                 # German (translated)
+    ```
+
+2. **Run refresh to generate and translate subtitles:**
+
+    ```bash
+    mkzforge refresh
+    ```
+
+    This will:
+
+    * Generate English subtitles using Whisper
+    * Automatically translate to Spanish, French, and German
+    * Preserve timing information from the original subtitles
+    * Create separate SRT files for each language in `build/`
+
+3. **Build your video with all subtitle tracks:**
+
+```bash
+mkzforge build
+```
+
+The final video will contain all subtitle tracks, properly mapped and labeled.
+
+### How Translation Works
+
+**Context-Aware Translation**: The entire transcript is translated as one document to preserve
+meaning, intent, and context. This produces much better results than translating line-by-line.
+
+**Timing Preservation**: After translation, the system intelligently splits the translated text to
+match the original subtitle timing, distributing words proportionally across subtitle entries.
+
+**Automatic Package Management**: Argos Translate language packages are downloaded and installed
+automatically on first use.
+
+### Example Output
+
+After running `mkzforge refresh` with multi-language support:
+
+```plain
+build/
+├── my-video.mkv              # Processed video
+├── my-video.txt              # Full English transcript
+├── my-video.en.srt           # English subtitles (Whisper)
+├── my-video.es.srt           # Spanish translation
+├── my-video.fr.srt           # French translation
+├── my-video.de.srt           # German translation
+└── my-video.mp4              # Final video with all subtitle tracks
+```
+
+### Supported Languages
+
+Any language pair supported by Argos Translate, including:
+
+* Spanish (es), French (fr), German (de), Portuguese (pt)
+* Italian (it), Russian (ru), Chinese (zh), Japanese (ja)
+* Arabic (ar), Hindi (hi), Korean (ko)
+* And many more!
+
+For detailed documentation, see [CHANGES.md](CHANGES.md).
+
+## @FutureFeature
+
+Features I'd like to add to this include:
+
+* `mkzforge publish` to publish to your configured social media platforms
+  * I want to support YouTube, TikTok, and Mastadon.
+  * Right now, only an SFTP endpoint is supported (uses [Fabric](https://www.fabfile.org/)).
+* Stream specifier error clarification: It would be nice to know that a stream is disconnected
+  right there in vscode or some editor of sorts. I hope a project like this can make that more
+  easily accesible as a potential by simulating the results of the configuration in ffmpeg's
+  `filter_complex` parameter.