Replicate Model Sync #215
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: Replicate Model Sync | |
| on: | |
| schedule: | |
| - cron: "45 */6 * * *" # Every 6 hours at minute 45 | |
| workflow_dispatch: # Allow manual trigger | |
| jobs: | |
| replicate-sync: | |
| runs-on: ubuntu-latest | |
| permissions: | |
| id-token: write | |
| contents: write | |
| pull-requests: write | |
| issues: write | |
| steps: | |
| - name: Checkout repository | |
| uses: actions/checkout@v4 | |
| with: | |
| fetch-depth: 0 | |
| - name: Install system dependencies | |
| run: | | |
| sudo apt-get update | |
| sudo apt-get install -y ffmpeg libgl1 libglib2.0-0 | |
| - name: Install uv | |
| uses: astral-sh/setup-uv@v4 | |
| with: | |
| enable-cache: true | |
| cache-dependency-glob: "uv.lock" | |
| - name: Install dependencies | |
| run: | | |
| uv sync --all-extras --dev --extra-index-url https://nodetool-ai.github.io/nodetool-registry/simple/ --index-strategy unsafe-best-match | |
| - name: Run Assistant | |
| uses: anthropics/claude-code-action@v1 | |
| env: | |
| ANTHROPIC_BASE_URL: https://api.z.ai/api/anthropic | |
| REPLICATE_API_TOKEN: ${{ secrets.REPLICATE_API_TOKEN }} | |
| MINIMAX_API_KEY: ${{ secrets.MINIMAX_API_KEY }} | |
| with: | |
| anthropic_api_key: ${{ secrets.ZAI_API_KEY }} | |
| additional_permissions: | | |
| actions: read | |
| claude_args: | | |
| --allowedTools "Bash,WebFetch,Edit,Read,Replace" | |
| prompt: | | |
| # Replicate Model Sync (Scheduled) | |
| You are an autonomous agent that adds new Replicate models to the **nodetool-replicate** repository. | |
| ## Step 0: Avoid Duplicate Work | |
| 1. Run `git branch -a` to check existing branches | |
| 2. If there's an open PR for similar work, continue on that branch instead of duplicating | |
| ## Step 1: Read Memory (Required) | |
| - `.github/opencode-memory/README.md` | |
| - `.github/opencode-memory/repository-context.md` | |
| - `.github/opencode-memory/node-creation-guide.md` | |
| - `.github/opencode-memory/features.md` | |
| ## Step 2: Discover New Models | |
| 1. Use the Replicate API or browse https://replicate.com/explore | |
| 2. Focus on: | |
| - Featured/Official models from major providers (OpenAI, Anthropic, Google, Meta, Stability, etc.) | |
| - Popular models with high run counts | |
| - Recent models (released in last 30 days) | |
| 3. Check `src/nodetool/nodes/replicate/gencode.py` to see which models already exist | |
| 4. Pick 1-3 new models that are NOT already in the `replicate_nodes` list | |
| ## Step 3: Add New Models | |
| 1. Edit `src/nodetool/nodes/replicate/gencode.py` | |
| 2. Add entries to the `replicate_nodes` list following the guide | |
| 3. Determine: | |
| - Correct `model_id` from Replicate URL | |
| - Appropriate `node_name` (PascalCase) | |
| - Correct `namespace` (image.generate, video.generate, etc.) | |
| - Correct `return_type` (ImageRef, VideoRef, AudioRef, str) | |
| - Any `overrides` for file input parameters | |
| ## Step 4: Regenerate Nodes | |
| ```bash | |
| # Generate the node Python files from gencode.py definitions | |
| python -m nodetool.nodes.replicate.gencode | |
| # Generate package metadata JSON (src/nodetool/package_metadata/) | |
| nodetool package scan | |
| # Generate DSL code (src/nodetool/dsl/) | |
| nodetool codegen | |
| ``` | |
| ## Step 5: Quality Checks | |
| ```bash | |
| ruff check . | |
| black --check . | |
| # Fix any issues before committing | |
| black . | |
| ``` | |
| ## Step 6: Update Memory | |
| Add entries to `.github/opencode-memory/features.md` documenting what was added. | |
| ## Guidelines | |
| - Add only 1-3 models per run to keep PRs focused | |
| - Prefer models from official/verified providers | |
| - Skip models that have no stable version | |
| - If a model fails to generate, skip it and try another | |
| - Always run linting before committing |