Loading ToolBench
Agent Context Protocol (ACP) installation and package management scripts for setting up ACP directory structures, managing packages, and tracking installations
This is a bash-based MCP server with 5 tools. Critical issues: (1) Transport is UNKNOWN, cannot verify MCP protocol compliance. (2) Tool definitions appear to be inferred from shell script filenames and comments rather than explicit MCP registration with JSON schemas. (3) Input schemas are partially visible for some tools (acp.package-create, acp.package-install, acp.package-list) but lack formal JSON Schema type definitions and validation. (4) Descriptions are present but generic and lack LLM-optimized guidance on WHEN to use each tool vs. alternatives. (5) No visible error handling, recovery guidance, or output schema documentation. (6) No security patterns (secret injection, permission gates, audit trails). (7) Tool composition is unclear, unclear if these are meant to be called by an LLM agent or are CLI utilities wrapped as MCP tools.
Agent Context Protocol (ACP) Installation Script - sets up the ACP directory structure and template files in a project
ACP Package Creator - creates a new ACP package with full ACP installation
Agent Context Protocol (ACP) Package Info Script - shows detailed information about an installed package
Agent Context Protocol (ACP) Package Install Script - installs third-party ACP packages (commands, patterns, designs, etc.) from git repositories
Agent Context Protocol (ACP) Package List Script - lists installed ACP packages with their versions and details
No explicit MCP tool registration visible. Tool definitions appear inferred from shell script filenames and comments rather than registered via MCP protocol with formal JSON schemas. Cannot verify tool names, descriptions, and schemas are actually exposed to MCP clients.
Input schemas lack formal JSON Schema type definitions. Parameters are documented as strings/booleans/arrays in comments, but no explicit 'type' field, 'required' array, or 'additionalProperties' constraints visible.
No output schemas documented. Tools return shell script output (text, YAML, or structured data) but no formal response schema is defined. LLMs cannot plan downstream tool calls or extract structured data without knowing response fields.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 31 | 2026-07-28+ | v2 |
Descriptions are generic and lack LLM-optimized guidance. E.g., 'Agent Context Protocol (ACP) Installation Script - sets up the ACP directory structure and template files in a project' (95 chars) does not explain WHEN to call this vs. acp.package-create, what prerequisites exist, or what happens on retry.
No error handling or recovery guidance. Shell scripts may fail (missing git, invalid URLs, permission errors) but no tool documentation explains what errors are retryable, what user-fixable, or what guidance to return to the LLM.