Educational workshop repository containing multiple MCP server implementations: Text Utility Server, Advanced Demo Server, and Secure Document Assistant
This is an educational workshop repository with 19 tools across multiple lab exercises. While tool names follow verb_noun conventions and descriptions are present, critical gaps prevent a higher score: (1) no input schemas are visible in the code, tools are decorated with @mcp.tool() but FastMCP auto-generates schemas from Python type hints, which are present but not validated in the source artifact; (2) descriptions are generic and brief (10-50 chars), lacking WHEN/WHY guidance and actionable context for LLM selection; (3) no output schemas documented; (4) no error handling guidance or recovery paths shown; (5) destructive operations (delete_document, remove_tag) lack confirmation patterns or dry-run support; (6) no parameter constraints (enums, ranges, regex) visible despite tools like 'process_data' accepting unbounded integers (1-100 stated in description but not enforced via schema). The lab starter files (Part_2) are intentionally incomplete skeletons with TODO comments and pass statements, so tool 1-4 have no real implementation visible. Solutions files (tools 5-19) show working Python implementations but schema definitions rely on implicit FastMCP introspection, which is not auditable from static source analysis alone.
Add a tag to a document.
Analyze text with multiple analysis stages.
Count the number of characters in the given text (including spaces).
Create a new document in the collection.
Delete a document from the collection.
Simulate fetching multiple URLs with progress.
Get statistics about the document collection.
No visible JSON Schema definitions in source code. Tools decorated with @mcp.tool() rely on FastMCP auto-generation from Python type hints. Without explicit schema validation, cannot confirm parameter types, ranges, enums, or required/optional status match the MCP spec requirements.
Descriptions are generic and under 50 characters for most tools (e.g., 'Count the number of words in the given text.' for word_count). They lack WHEN/WHY context, LLM-selection guidance, and actionable details.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 45 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 26 | - | v1 |
Count the number of lines in the given text.
List all available documents with their metadata.
List all tags and their document counts.
Process a batch of items with progress reporting.
Remove a tag from a document.
Return the input text reversed.
Find all documents with a specific tag.
Search for documents containing the specified text.
Convert the given text to title case.
Convert the given text to lowercase.
Convert all characters in the text to uppercase.
Count the number of words in the given text.
Destructive operation 'delete_document' has no confirmation pattern, dry-run support, or recovery guidance. Per pattern:confirmation-request, irreversible operations should support confirm-before-execute to prevent agent mistakes.
process_data accepts 'items' parameter with stated constraint '1-100' in description only, not enforced in schema or validation code. Per review:param-validation-rules, numeric ranges must be specified in both description AND schema (min/max), with early validation and actionable error messages.
No error handling or recovery guidance visible in tool implementations. Tools return generic success messages or assume success. Per pattern:recovery-guide, errors must tell the LLM what to do next (retry, user-fixable, unrecoverable classification).
Output schemas not documented. No indication of what fields are returned, their types, or structure. Per pattern:tool, LLMs need documented return types to plan downstream calls and extract data correctly.
Starter files (tools 1-4) contain TODO comments and pass statements, no executable implementation. Per HARD SCORING RULES, tools without visible definitions are capped at 50 overall.
No input validation or constraint enforcement visible. Tools like search_documents accept free-form query strings with no regex, length limits, or character restrictions stated. Per review:param-validation-rules, constraints must be explicit in descriptions and enforced at runtime.
Tools list_documents and search_documents have no pagination support (no limit, offset, or next_cursor parameters). Per pattern:paginated-result, tools returning lists should accept pagination parameters and return total counts to avoid context window overflow.