Loading ToolBench
MCP server providing tools for web research, file operations, command execution, markdown-to-PDF conversion, and persistent SQLite-based knowledge storage for project tracking and issue management
Server has 31 tools with complete input schemas and descriptions visible in tools.json. Naming follows verb_noun convention (add, today, web_search, list_files, read_file, write_file, delete_file, etc.). However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool vs alternatives. Parameter descriptions exist but are minimal (avg ~40 chars). No output schemas documented. Error handling is basic. Security concerns: run_command and file operations expose path traversal risks without clear sanitization guidance. Destructive tools (delete_file, remove_directory, clear_context) lack confirmation patterns. Overall: functional but below production baseline.
Adds two numbers together.
Add a detailed step to a project change record.
Add an alias mapping for a project to normalize different naming conventions.
Add a project change record with type and summary.
Append text content to the end of an existing file without overwriting.
Clear all context entries for a specific project.
Delete a file from the filesystem.
Scrape a URL, extract main content, and return a clean text summary.
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields. Baseline: 100% of A+ tools document return types.
Destructive tools (delete_file, remove_directory, clear_context) lack confirmation or dry-run patterns. Agents can cause irreversible data loss without safeguards.
run_command accepts arbitrary shell commands with no input validation or sanitization guidance. Path traversal and command injection risks are not mitigated.
File operation tools (read_file, write_file, list_files) accept relative paths without explicit path traversal protection documented. LLMs could be tricked into accessing files outside sandbox.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | <=2025-11-25 | v2 |
Get detailed history of a specific project change.
Get detailed information about a specific issue.
Insert text after a specific marker line in a file.
List files and directories in a specified folder.
List all issues for a specific project.
List all changes for a specific project.
List all projects in the knowledge store.
Convert a Markdown file to a styled PDF document using WeasyPrint.
Query the SQLite knowledge store using full-text search.
Query issues using full-text search.
Read a file with AI-friendly line-based windowing. Returns formatted output with navigation hints for paging through large files.
Remove a directory and all its contents recursively.
Replace text in a file using regex or literal string matching.
Execute a shell command within the sandbox directory.
Search for files matching a pattern in a directory.
Search project changes using full-text search.
Store a key-value pair in the SQLite knowledge store for a specific project.
Store an issue record in the knowledge store.
Get today's date and time.
Update the project association of an issue.
Update the status of an issue.
Search the live web for current information, news, or facts.
Write text content to a file. Creates parent directories if needed.
Descriptions are generic and lack LLM-optimized guidance. No WHEN-to-use context, no dependency hints, no distinction between similar tools (e.g. query_context vs search_project_changes). Baseline: 50-200 chars with clear intent.