Standalone TypeScript MCP server for filesystem access
Strong schema definitions with complete input parameters and clear descriptions. All 16 tools have explicit schemas with type definitions and parameter descriptions. Tool names follow verb_noun convention (read_file, write_file, search_files). Descriptions average ~120 chars, meeting the 10-1024 char baseline. However, descriptions lack actionable context (when to use vs similar tools, prerequisites, recovery guidance). Output schemas are not documented. Error handling is present but lacks recovery guidance. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels (READ_ONLY, WRITE, DESTRUCTIVE). Missing dependency hints and chaining IDs in responses.
Copy a file or directory to a new location.
Create a new directory. Parent directories can be created as needed.
Delete a file or directory. Directories can be deleted recursively.
Edit a file by finding and replacing text patterns. Supports substring and regex matching with optional context display.
Get detailed metadata about a file or directory including size, timestamps, and permissions.
Read the first N lines of a file.
List the directories and files that are accessible to this MCP server.
Output schemas not documented. LLMs cannot infer what fields to expect from tool responses, forcing them to guess at downstream field names and risk broken tool chains.
Tool annotations missing. Despite clear risk levels (READ_ONLY, WRITE, DESTRUCTIVE), tools lack readOnlyHint, destructiveHint, and idempotentHint annotations. Agents cannot distinguish safe from dangerous operations without reading descriptions.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 63 | 2025-06-18+ | v2 |
List the contents of a directory, optionally recursively. Returns file and directory entries with metadata.
Move or rename a file or directory to a new location.
Read the complete contents of a file. Returns the file content as text, along with metadata like encoding, size, and line count.
Read a file and return its contents as base64-encoded data. Useful for binary files.
Read multiple files in a single request. Returns an array of file contents with metadata for each file.
Search for text patterns within files. Supports substring and regex matching with optional context lines.
Search for files by name using glob patterns. Returns matching file paths.
Read the last N lines of a file.
Create a new file or overwrite an existing file with the provided content. Supports atomic writes with optional hash/mtime validation.
Descriptions lack actionable context. Descriptions state WHAT tools do but not WHEN to use them vs similar tools (e.g., search_files vs search_file_content), prerequisites, or recovery guidance. Average description length ~120 chars is below the 194-char production baseline.
No error recovery guidance. Error responses (inferred from code) do not tell LLMs what to do next. A 'file not found' error should suggest 'Try search_files() to locate the file' or 'Check list_allowed_directories() for accessible paths.'
Destructive operations lack confirmation step. delete_file with recursive=true can destroy entire directory trees. No dry-run or confirmation mechanism to prevent catastrophic agent mistakes.