Loading ToolBench
Toolbelt exhibits significant definition quality gaps across all 11 tools. While tool names follow verb_noun conventions (generate_unit_tests, sanitize_n8n_workflow), descriptions are present but lack LLM-optimization guidance. Critical issue: NO input schemas are visible in the provided source code. Only two tool implementations are shown (bootstrapper.py, packer.js); the remaining 9 tools are inferred from the tool list without explicit schema definitions. Parameter descriptions exist in code comments but are not formally structured. Error handling is minimal, most tools exit with generic error messages rather than recovery guidance. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) detected.
Analyze SQL schema definitions (DDL) and generate TypeScript type definitions and Supabase Row Level Security (RLS) policies
Audit ESG reports for greenwashing claims, framework alignment (GRI, SASB, CSRD, TCFD), and calculate Greenwashing Risk Score (GRS) and Disclosure Completeness Index (DCI)
Calculate Scope 1, Scope 2, and Scope 3 greenhouse gas emissions in metric tons of CO2 equivalent (tCO2e) based on IPCC AR6 GWP100 and GHG Protocol standards
Scan file for AI-slop patterns (delve, tapestry, testament, moreover, furthermore, bespoke, revolutionize, pivotal, throat-clearing intros) and report occurrences
Sanitize file by removing AI-slop jargon using Gemini API, rewriting text to be direct and human-like
Compact codebase directory by extracting and summarizing Python and JavaScript/TypeScript file structures for token optimization
No input schemas visible in source code. Only two tool implementations shown; 9 tools inferred from list without explicit registration.
Descriptions lack LLM-optimization. Most are 50 - 65 chars, below the 194-char baseline for A+ tools. No WHEN-to-use guidance, no dependency hints, no recovery paths.
Error handling is generic. bootstrapper.py exits with 'Error calling Gemini API: {e}' and 'GEMINI_API_KEY not defined', no recovery guidance, no actionable next steps for LLM.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 33 | 2026-07-28+ | v2 |
Generate valid JSON-LD SEO schema markup based on business information using Gemini
Analyze git diff and generate conventional commit message and professional Pull Request description using Gemini
Analyze source code file and generate a robust, production-grade unit test suite using Gemini
Fetch top stories from Hacker News and generate curated markdown digest using Gemini
Sanitize n8n workflow credentials and secrets by redacting sensitive keys, webhook URLs, and credential references
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Tools marked WRITE risk (generate_unit_tests, sanitize_n8n_workflow, analyze_sql_schema, generate_pr_info, generate_json_ld_schema) lack destructiveHint annotation.
Hardcoded API key fallback in bootstrapper.py (C:\Users\araga\.hermes\.env) is a security anti-pattern. Credentials must use server-side secret injection, never embedded in tool code.