Forge AI Skill

Interacts with the Forge AI platform API using src/forge-skill/forge_client.py.

Quick Reference

python3 src/forge-skill/forge_client.py \x3Ccommand> [args]

Commands

Authentication

Articles

Evaluations

Tags

JSON File = HTTP Body

The JSON file is the exact data field sent in the API request body.

Article Create (article-crud)

{
  "title": "My Article Title",
  "content": "# Content in markdown...",
  "tags": ["tag_id_1", "tag_id_2"],
  "coverImage": "https://example.com/cover.jpg",
  "type": 0
}

Article Update (article-crud)

Add articleId with the pending wrapper:

{
  "articleId": "art_xxx",
  "pending": {
    "title": "Updated Title",
    "content": "# Updated content...",
    "tags": ["tag_id_1"],
    "coverImage": "https://example.com/new-cover.jpg"
  }
}

Evaluation Create (evaluation-crud)

{
  "pending": {
    "modelName": "GPT-4",
    "skillName": "Code Generation",
    "title": "Evaluation Title",
    "content": "# Evaluation content...",
    "modelVersion": "2024-05",
    "skillDescription": "Generate production code",
    "skillTags": "python, go",
    "overallScore": 85,
    "dimensions": {
      "准确性": 90,
      "推理能力": 80
    },
    "remark": "Good overall performance"
  }
}

Evaluation Update (evaluation-crud)

Add evaluationId:

{
  "evaluationId": "ev_xxx",
  "pending": {
    "modelName": "GPT-4",
    "skillName": "Code Generation",
    "title": "Updated Evaluation",
    "content": "# Updated content..."
  }
}

State

Typical Workflow

1. forge_client.py login [email protected] mypassword
2. Write .json file matching the HTTP body structure
3. forge_client.py article create .forgeai/articles/my-article.json
4. Take the returned ID, add it to the JSON file
5. Edit content, run forge_client.py article update .forgeai/articles/my-article.json