model-discovery
Fetch current model names from AI providers (Anthropic, OpenAI, Gemini, Ollama), classify them into tiers (fast/default/heavy), and detect new models. Use when needing up-to-date model IDs for API calls or when other skills reference model names.
Оригинален текст · няма потвърден превод на българскиПърво прочетете инструкцията.
Файлът се зарежда от посочения източник само по ваше действие. Можете да го прочетете тук, да го копирате или да го запазите като .md.
GitHub хранилище- Звезди на репото
- 428
- Лиценз
- Не е посочен. Проверете в репозитория.
- Последна проверка на записа
- Няма посочена дата
Добавете в своя инструмент.
Командата отваря избора на skills от репозитория. Прочетете файла и условията на автора преди изпълнение.
npx skills add aiskillstore/marketplaceСтраницата копира текст. Не инсталира и не изпълнява нищо на устройството ви.
Откъс от SKILL.md
Началото на файла на автора, без превод.
Model Discovery Skill
Fetch the most recent model names from AI providers using their APIs. Includes tier classification (fast/default/heavy) for routing decisions and automatic detection of new models.
Variables
| Variable | Default | Description |
|---|---|---|
| CACHE_TTL_HOURS | 24 | How long to cache model lists before refreshing |
| ENABLED_ANTHROPIC | true | Fetch Claude models from Anthropic API |
| ENABLED_OPENAI | true | Fetch GPT models from OpenAI API |
| ENABLED_GEMINI | true | Fetch Gemini models from Google API |
| ENABLED_OLLAMA | true | Fetch local models from Ollama |
| OLLAMA_HOST | http://localhost:11434 | Ollama API endpoint |
| AUTO_CLASSIFY | true | Auto-classify new models using pattern matching |
Instructions
MANDATORY - Follow the Workflow steps below in order. Do not skip steps.
- Before referencing model names in any skill, check if fresh data exists
- Use tier mappings to select appropriate models (fast for speed, heavy for capability)
- Check for new models periodically and classify them
Red Flags - STOP and Reconsider
If you're about to:
- Hardcode a model version like
gpt-5.2orclaude-sonnet-4-5 - Use model names from memory without checking current availability
- Call APIs without checking if API keys are configured
- Skip new model classification when prompted
STOP -> Read the appropriate cookbook file -> Use the fetch script
Workflow
Fetching Models
- Determine which provider(s) you need models from
- Check if cached model list exists:
cache/models.json - If cache is fresh (< CACHE_TTL_HOURS old), use cached data
- If stale/missing, run:
uv run python scripts/fetch_models.py --force - CHECKPOINT: Verify no API errors in output
- Use the model IDs as needed
Checking for New Models
- Run:
uv run python scripts/check_new_models.py --json - If new models found, review the output
- For auto-classification:
uv run python scripts/check_new_models.py --auto - For interactive classification:
uv run python scripts/check_new_models.py - CHECKPOINT: All models assigned to tiers (fast/default/heavy)