The Tradeics Docs skill teaches agents to navigate the official developer hub, prefer canonical URLs, and refuse invented API routes.
Goals
- Always ground answers in Tradeics docs (
developers.tradeics.com). - Prefer
/llms.txt→ relevant guide/reference → cite the path. - Distinguish Guides (concepts) from References (HTTP contracts).
- Pair with Docs MCP for live search when available.
Behavior contract (fill package with this)
| Rule | Agent must |
|---|---|
| Source of truth | Use official docs; never invent /v1/ REST routes |
| Discovery | Start from llms.txt, deepen with llms-full.txt only when needed |
| Citations | Link concrete guide/reference paths |
| Secrets | Never ask users to paste live tokens into shared skills |
| Placeholders | Keep .example hosts until production domains are confirmed |
Contents checklist (for skill authors)
Use this checklist when packaging the skill — structure is fixed; body copy can expand:
- Identity — name, description, when to trigger.
- Primary URLs — hub base,
llms.txt, REST V2, LLM, MCP catalog. - Map of the hub — Getting Started, products, Agent Skills, MCP Servers, CLI, References.
- Citation template — how to quote paths and titles.
- Refusal rules — no undocumented private endpoints; no secret dumping.
Example prompts
- “Find how Tradeics handles pagination and link the guide.”
- “Compare REST V2 vs MCP for catalog sync — cite both.”
- “Give me the Getting Started path for webhooks.”