Loading ToolBench

Reading the current corpus…

claude-skill-debug MCP server · F · ToolBench

claude-skill-debug

A debugging skill for Claude Code projects that helps identify and clean up debug markers, manage test environments, and prevent accidental commits of debug code

2026-07-28+3 tools

30/100

Definition Quality38

Protocol Readiness0

Supportability36

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence

Current-spec patterns detected

Summary

This is a bash-based skill installer, not an MCP server. The three tools (check-residual, cleanup, init-test-env) are shell scripts invoked via a Claude Code skill framework, not MCP protocol tools. Tool definitions are inferred from script filenames and comments rather than explicit MCP registration. Schemas are partially present but minimal. Descriptions exist but lack LLM-optimization guidance. No error recovery hints, no output schema documentation, no tool annotations. The server does not implement MCP protocol at all, it's a local skill installer.

Tools (3)

check-residualread only50/100

Check for AI-DEBUG residual markers in project source code. Supports --pre-commit mode to intercept git commits and prevent accidental commits of debug code

cleanupwrite50/100

Clean up all AI-DEBUG marked debug code from project source files. Optionally filters by session ID to remove only specific debug markers

init-test-envwrite43/100

Create test accounts and save credentials to test configuration file. Registers test users with the backend API and retrieves authentication tokens

Issues (5)

CRITICALtool ↗

Not an MCP server, bash skill installer for Claude Code. Tools are shell scripts, not MCP protocol tools. No MCP transport, no tool registration via MCP, no protocol compliance.

check-residualcleanupinit-test-env

CRITICALresponse-shaper ↗

Output schemas completely undocumented. check-residual returns JSON hookSpecificOutput on --pre-commit mode but no schema declared. cleanup and init-test-env return shell exit codes and stdout text with no structured output definition.

check-residualcleanupinit-test-env

HIGHrecovery-guide ↗

Error handling lacks recovery guidance. Scripts exit with code 1 on failure (e.g., 'Backend not running at $API_BASE') but provide no actionable next steps for the LLM. No error classification (retryable vs fatal).

init-test-envcheck-residual

HIGHconstrained-input ↗

Recommendations

Convert to a proper MCP server (Node.js, Python, or Go) that implements the MCP protocol over HTTP/SSE or stdio. Register tools via MCP's tool definition mechanism with full JSON Schema input/output.
Document output schemas for all three tools. Define structured return types (e.g., check-residual returns {status: 'clean'|'found', markers: [{file, line, count}], message: string}).
Add error recovery hints to all error paths. E.g., 'Backend not running at $API_BASE. Ensure the API server is running on port 3000, or set TEST_API_BASE environment variable.'
Enhance parameter descriptions with format constraints. E.g., api_base: 'HTTP(S) URL of backend API (e.g., http://localhost:3000 or https://api.example.com). Required.'
Move hardcoded credentials to environment variables or a secure vault. Replace default passwords with a note: 'Credentials must be set via TEST_USER_A, TEST_PASS_A, TEST_USER_B, TEST_PASS_B environment variables for security.'

Score history

Overall score trend

First recorded score · v2 rubric

30/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	F	30	2026-07-28+	v2