Dual-protocol AI portfolio server combining MCP (FastMCP) for structured tool access and A2A agent for conversational discovery. Exposes Mario's professional background, skills, projects, services, and contact information through two interfaces.
This portfolio MCP server exposes 7 read-only tools via fastmcp (HTTP transport). All tools have descriptions and basic parameter schemas. However, several gaps prevent a higher score: (1) descriptions lack sufficient detail for LLM selection, many are generic summaries without explaining when to use them or what context they enable; (2) parameters have minimal documentation, the 'category' enum in get_skills is the only parameter with meaningful constraint; (3) no output schemas are documented, forcing LLMs to guess response structure; (4) error handling is absent, no guidance on recovery or edge cases; (5) tool composition could be tighter, no indication of how results chain (e.g., get_projects lists projects, then get_project_detail requires a 'project_name', but there's no explicit mapping). Naming is solid (all verb_noun, clear intent). The server is a reasonable portfolio showcase but lacks production-grade tool engineering.
Get Mario's professional bio, background, timezone, and availability.
Get contact information, availability, and how to hire Mario.
Get Mario's professional experience timeline and education.
Get a detailed deep-dive on a specific portfolio project. Args: project_name: The project name or slug. Use get_projects() first to see available projects.
Get a summary list of all portfolio projects.
Get the services Mario offers and his pricing approach.
Get Mario's technical skills, optionally filtered by category. Args: category: Filter by category. Options: 'ai', 'automation', 'backend', 'frontend', 'all'.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract fields without inferring response structure from examples or trial-and-error.
Descriptions are generic summaries (10 - 45 chars) lacking LLM guidance. 'Get Mario's professional bio, background, timezone, and availability' does not explain when to call get_about_me vs get_contact_info, or how to compose results. Descriptions should be 50 - 200 chars and answer: What does it do? When should I call it? What context does it provide?
Parameter 'project_name' in get_project_detail lacks constraint documentation. Description says 'Use get_projects() first to see available projects' but does not document the expected format (slug, full name, ID?) or provide an enum. This forces LLMs to guess.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 52 | <=2025-11-25 | v2 |
| 2026-03-09 | C | 63 | - | v1 |
No error handling guidance. If get_project_detail is called with an invalid project_name, what happens? Does the tool return an empty result, a 404, or an error message? Without error classification (retryable, user-fixable, fatal), LLMs cannot recover gracefully.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). While all tools here are read-only and safe, explicitly declaring this via annotations helps LLMs reason about side effects and retry safety.