Loading ToolBench
A collection of skills, scripts, and utilities for Claude Code development. Includes todo app examples, Loki Mode autonomous wrapper, session management, validation checklists, and various development utilities.
This server has critical definition gaps across all 6 tools. Most tool descriptions are present but lack actionable detail for LLM selection. Input schemas are visible but incomplete: parameters lack type definitions in several tools (e.g., loki_wrapper has only 'prd_path' with no type), and descriptions are generic or missing. No output schemas are documented. Error handling is absent. Tool naming is reasonable (verb-noun style) but composition issues exist: 'checklist' and 'session_manager' are discovery/orchestration tools that don't fit the agent pattern. The server is STDIO-only, which is a hard transport cap.
Manages (start/stop/status) the local development server for previewing the application. Detects dev/start scripts from package.json and manages server lifecycle with PID tracking.
Master checklist runner that orchestrates all validation scripts in priority order. Runs security scans, lint checks, schema validation, tests, UX audits, SEO checks, and optional performance checks.
Generates a rich index of skills from SKILL.md files in the skills directory, parsing YAML frontmatter and extracting metadata including name, description, risk level, and source information.
Autonomous wrapper script that launches Claude Code with Loki Mode prompt, monitors the process, handles rate limits with exponential backoff, and automatically resumes from checkpoints on interruption.
Analyzes project state, detects tech stack from package.json, tracks file statistics, and provides a summary of the current session including detected modules and features.
Starts one or more servers, waits for them to be ready on specified ports, runs a command, then cleans up all servers. Supports multiple concurrent servers with port polling.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields. Violates pattern:tool and pattern:response-shaper.
loki_wrapper schema incomplete: 'prd_path' parameter lacks explicit type definition.
checklist and session_manager are orchestration/discovery tools, not agent-composable actions. They violate pattern:tool (each tool should do exactly one thing). checklist runs 'all validation scripts in priority order', this is a workflow, not a tool.
auto_preview and loki_wrapper descriptions lack actionable detail. 'Manages (start/stop/status) the local development server' does not explain WHEN to call it vs other tools or what it returns. Descriptions under 100 chars risk LLM confusion.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | 2026-07-28+ | v2 |
No error handling guidance in any tool. If a server fails to start (with_server) or a skill index is malformed (generate_index), the LLM has no recovery path. Violates pattern:recovery-guide.