AI Skill 檔案
將任一 Skill URL 貼入您的 AI 助理(Claude、Cursor、Copilot、ChatGPT…),即可讓它立即掌握該 BazaarLink 主題的完整知識。無需設定,助理讀取檔案後即可上手。
- 複製下方 Skill URL
- 開啟您的 AI 助理
- 告訴它:「載入這個 skill:<URL>」
- 助理即可掌握該主題的完整知識
What Are Skills?
Skills are pre-built, reusable prompt workflows that run inside Claude Code and other agentic environments. Each skill encapsulates a specific capability — from writing implementation plans to reviewing pull requests — letting your AI agent invoke complex multi-step behaviors with a single command.
BazaarLink hosts a growing library of skills. You can browse available skills, call them via the API, or build your own. Skills run server-side through the BazaarLink Worker runtime, keeping your prompts fast and your context window clean.
How to Call a Skill
Call any skill through the BazaarLink API using the standard completions endpoint. Pass the skill name as a system prompt prefix or use the X-Skill header for automatic routing.
import openai
client = openai.OpenAI(
base_url="https://bazaarlink.ai/api/v1",
api_key="YOUR_API_KEY"
)
response = client.chat.completions.create(
model="claude-sonnet-4-6",
messages=[{"role": "user", "content": "Review this PR: #123"}],
extra_headers={"X-Skill": "pr-review"}
)
print(response.choices[0].message.content)Skills vs. Models
Models are the raw intelligence — they generate text, code, and structured data. Skills are workflows built on top of models: they provide the system prompt, enforce a reasoning structure, and return results in a predictable format. Think of skills as the domain expertise layer and models as the execution engine underneath.
When you call a skill, BazaarLink automatically selects the best underlying model for that task and injects the skill's curated system prompt. You get consistent, high-quality output without managing prompt engineering yourself.
Skills
BazaarLink API 完整指南:認證、模型 ID、Chat Completions、Embeddings、串流、Tool Calling、路由、金鑰管理及 Agent 自動註冊。
透過 Bearer Management Key 程式化管理組織、Team 與成員。涵蓋所有 /api/v1/orgs/ 端點,附 curl、Python、TypeScript 範例。