Loading ToolBench
This server presents a satirical/humorous HR tool suite with serious deficiencies in production readiness. All four tools have names that violate verb-noun conventions (fire_my_manager, increase_salary, request_unpaid_overtime, generate_corporate_jargon). While descriptions exist and schemas are partially visible via Zod, the tools collectively mock core HR functions rather than perform them legitimately. Schemas show reasonable parameter validation (e.g., percentage bounds 0-100, hours 1-24), but responses are hard-coded comedic strings with no actual integration to HR systems, data persistence, or side effects. This is not a production tool, it's a joke implementation. Scoring reflects the fundamental uselessness for any real agent workflow, despite technically present (if incomplete) schema definitions.
Submit a formal complaint about your manager that may lead to termination
Create meaningless buzzword-filled content to impress management
Increase your salary by a specified percentage or amount
Request to work extra hours for free to demonstrate your commitment
Tool names do not follow verb_noun convention. 'fire_my_manager' is vague (does it submit a complaint, terminate employment, or log a grievance?). Should be 'submit_manager_complaint' or 'file_misconduct_report' with clear action verbs.
Tools return hard-coded comedic strings (e.g., 'Your request will be thoroughly ignored by HR for 12 business weeks') with no actual state mutation, persistence, or integration to backend systems. An agent cannot rely on these tools to perform actual HR operations, they are mock implementations.
Output schemas are not documented. Tools return unstructured free-text strings. LLMs cannot parse structured fields or chain results to downstream tools. Responses should return {status, request_id, estimated_days, approval_probability} instead of markdown-formatted comedy text.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 41 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 19 | - | v1 |
Tool descriptions are generic and do not explain WHEN to use each tool or what constitutes success. 'Increase your salary by a specified percentage or amount' lacks context: is this for salary negotiation, promotion, or cost-of-living adjustment? Descriptions should answer: When would an agent call this? What are the prerequisites? What does success look like?
Parameter descriptions lack actionable format/constraint guidance. 'Reason for requesting a salary increase' is vague, should it be 10-500 characters? Must it cite performance metrics? Descriptions should state: 'Free-form text (50-300 chars) explaining business justification (performance improvement, market rate, promotion).'
No error recovery guidance. If increase_salary fails (e.g., reason is empty), the error is 'Reason for requesting a salary increase is required' with no hint about what makes a valid reason or what to do next. Errors should be: 'Reason cannot be empty. Provide a specific business justification (e.g., performance improvement, market rate alignment).'
Tool semantics are contradictory or satirical, not functional. 'request_unpaid_overtime' is framed as celebrating free labor, no legitimate business would expose this as a public tool. 'fire_my_manager' is not a complaint submission; firing is a management action, not an employee action. Tool design does not align with real HR workflows.