| name | mcp-research-orchestrator |
| description | Produce a source-backed research package via MCP tools (Context7, Microsoft Learn, GitHub, Terraform registry) before writing or updating a technical article for this site. Use when creating a new technical article, updating one after a stack change, or making a site-improvement recommendation that depends on external/current reference material. |
| metadata | {"ported-from":"ai/skills/mcp-research-orchestrator.SKILL.md"} |
MCP Research Orchestrator
Adapted from this repo's ai/ context kit. Read ai/mcp/source-of-truth-policy.md and src/data/articles/types.ts (the Article interface) before starting — every claim produced here should end up traceable to a source, and the output should slot directly into an Article.content.
When to use
- Creating a new technical article in
src/data/articles/artigos/
- Updating an existing article after a stack/version change
- Any site-improvement recommendation that depends on external reference material rather than repo state
Required inputs
- topic: specific, not vague — "Azure Container Apps with Dapr" is usable, "cloud" alone is not
- target audience: technical level + role (e.g. "mid-level platform engineer")
- optional: depth (beginner/intermediate/advanced, default intermediate), scope (conceptual/tutorial/hands-on/comparative)
Reject a vague topic before researching — ask for a narrower one instead of guessing.
Research steps (use the actual MCP tool, fully qualified by server name)
- Library/product docs — resolve the library, pull current API/examples/breaking changes via the Context7 MCP tools available in this session.
- Microsoft Learn — confirm official guidance and architecture recommendations, extract security/operational best practices, via the Microsoft Learn MCP tools.
- GitHub — search real, active code for implementation patterns and pitfalls.
- Terraform registry — verify provider/resource names, parameters, and current versions if the article touches IaC.
- Consolidate — write an objective technical summary, separate facts (sourced) from opinion, and confirm every claim has a source before handing off to writing.
Output shape
- Executive summary (5-8 lines)
- Evidence grouped by source (library docs / Microsoft Learn / GitHub / Terraform registry)
- Top 5 pitfalls to avoid
- Final recommendation, ready to become article content per
publishing-content
