MCP server for storing, loading, and searching files with keyword matching capabilities
The server defines 5 tools with explicit schemas and basic descriptions, but falls short of production quality. Tool naming is clear and action-oriented (store_, load_, search_, list_), which is good. However, parameter descriptions are missing or minimal across almost all tools, and output schemas are undocumented. Critical issues: (1) fileName parameter in store_file, load_file, and search_keyword lacks descriptions; (2) keyword parameter in search_keyword has no description; (3) content parameter in store_file has a description but fileName does not; (4) none of the tools document their return structure or response format, making it impossible for LLMs to plan follow-up actions; (5) no error handling guidance, load_file returns error text but doesn't help the LLM self-correct; (6) store_file and load_file have side effects (writes to memory, reads from disk) but descriptions do not clearly state these are state-modifying operations; (7) search_keyword's output (matching lines or count?) is undefined. The server lacks output schema documentation entirely, which violates the pattern:tool requirement. Average tool quality is fair but below B-grade threshold due to undescribed parameters and missing response documentation.
List all files in the server
List all files that are currently loaded/stored in memory
Loads a file from the directory in memory for searching
Search for keyword within a stored file
Stores a file directly with its content in memory for searching
Parameter descriptions missing or incomplete. fileName appears in 3 tools (store_file, load_file, search_keyword) with zero, zero, and zero descriptions respectively. keyword in search_keyword has no description. Per the rubric, every parameter must have a non-empty description explaining what it controls.
Output schemas completely undocumented. store_file, load_file, search_keyword, list_directory_files, and list_loaded_files return text responses but the structure (what fields? what format? what does search_keyword return, matching lines, count, or full content?) is undefined. LLMs cannot plan downstream calls or extract required data without documented return types.
State-modifying operations (store_file, load_file) do not declare side effects in descriptions. Descriptions must state WHAT the tool does, including whether it modifies state. Users and LLMs need to know which calls are safe to retry and which have irreversible consequences.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 44 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 43 | - | v1 |
Error handling provides no recovery guidance. load_file catches read errors and returns error text, but does not tell the LLM what to do next (e.g., 'File not found. Try list_directory_files() to see available files.').
search_keyword behavior undefined. Does it return all matching lines? Just a count? Full file content with highlights? Without documented output structure, LLMs cannot reliably use this tool or chain it to others.
list_directory_files and list_loaded_files lack descriptions explaining the distinction or what structure they return. Are they returning a simple list of names or structured metadata? Do they support pagination? Completely unclear.
No input validation hints. If a fileName parameter expects specific characters or patterns (e.g., no path separators to prevent directory traversal), or if file size is limited, these constraints must be stated in parameter descriptions. Currently absent.