Loading ToolBench
A monorepo-based agent platform with chat, workspace, MCP server integration, and skill management capabilities
Agent-x provides 13 tools with mostly complete schemas and descriptions. Naming follows verb_noun conventions (listSkills, loadSkill, createFile, readFile, etc.). Descriptions are present and contextual (avg ~120 chars), explaining WHAT each tool does and WHEN to use it. However, several parameters lack descriptions (e.g., getCurrentTime has no visible schema/description), and output schemas are not documented in the source. Error handling guidance is absent, tools do not indicate what to do on failure (e.g., 'file not found', 'skill not found'). Workspace tools properly validate paths against '..' and absolute paths, but no recovery hints are provided. Composition is strong: tools chain well (listSkills → loadSkill, listFiles → readFile), and batch operations (writeFiles) reduce sequential calls. Risk annotations (READ_ONLY, WRITE, DESTRUCTIVE) are present but not exposed in tool definitions for LLM visibility.
Create a new file in the workspace. For text files, provide content as a string. For binary files, provide base64-encoded content.
Delete a file from the workspace.
Check if a file or directory exists in the workspace. This is a lightweight check that does not read file content.
Get file metadata (size, MIME type, line count for text files) without reading the full content. Use this instead of readFile when you only need to check file properties.
List all files in the workspace or in a specific subdirectory.
List all skills configured for this agent. Returns each skill's ID, name, and description. Use loadSkill to read the full content of a specific skill.
Load the full content of a skill by its ID.
getCurrentTime tool has no visible description or schema in source code. Cannot verify parameter requirements or output format.
Output schemas are not documented. LLMs cannot infer what fields listSkills, readFile, fileInfo, or searchFiles return, forcing them to guess downstream field names and risk broken tool chains.
No error recovery guidance. Tools do not indicate what to do if a file is not found, a skill ID is invalid, or a path traversal attempt is detected. LLMs receive no actionable next steps.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
Read a file from the workspace. Returns text content for text files and base64-encoded content for binary files.
Read specific lines from a text file in the workspace. Use this instead of readFile when you only need to read a portion of a large file. Line numbers are 1-based.
Search for text across workspace files (case-insensitive). Returns matching lines with line numbers. Optionally restrict search to a subdirectory.
Update the content of an existing file in the workspace.
Write multiple files to the workspace atomically. All files are written or none (atomic batch operation).
Destructive operations (deleteFile) lack confirmation or dry-run support. An agent mistake could permanently delete files without a chance to undo.
Risk annotations (READ_ONLY, WRITE, DESTRUCTIVE) are present in metadata but not exposed in tool definitions. LLMs cannot see which tools are safe to retry and which have irreversible side effects.