Deterministic equity-compensation calculator tools for AI agents. Multi-year ISO/AMT optimization, NSO sell-vs-hold, RSU vest decisions, single-stock concentration, protective put pricing, QSBS qualification.
Critical failure: Cannot verify ANY tool definitions in source code. The repository provides NO visible tool schemas, parameter definitions, or descriptions in the supplied code excerpt. Tool names exist in config files and package.json, but actual tool registration code (which should define schemas, parameter types, descriptions, and input validation) is absent from the provided source. The Dockerfile references 'src/stdio-server.ts' but that file's content is not included. Without access to the actual tool implementation, all schemas and detailed descriptions remain unverifiable. The brief descriptions in the TOOLS list ('Multi-year ISO/AMT optimization calculator', etc.) are generic and under 50 characters, well below the 194-char baseline for production tools. No parameter schemas, input validation, or error handling guidance visible. This violates multiple hard rules: cannot confirm schemas exist, descriptions are too short, and tool registration code is not in the supplied source.
Protective collar (put/call) pricing and hedging calculator
Single-stock concentration analysis and optimization
Equity-funding plan calculator
Multi-year ISO/AMT optimization calculator
NSO sell-vs-hold decision calculator
Protective put pricing and hedging calculator
QSBS (Qualified Small Business Stock) qualification and tax analysis
RSU vest decision calculator
Tool definitions not provided in source code. No input schemas, parameter types, or detailed descriptions visible. Cannot verify compliance with DEFINITION QUALITY rubric.
Tool descriptions are under 50 characters (e.g., 'Multi-year ISO/AMT optimization calculator', 'NSO sell-vs-hold decision calculator').
No parameter schemas visible in supplied code. Cannot confirm tools have input_required fields with type definitions, constraints, descriptions, or enum values.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 31 | 2026-07-28+ | v2 |
No output schemas documented. Cannot verify that tools return structured objects with typed fields, include chaining IDs (e.g., next_cursor for paginated results), or provide proper error guidance.
Tool names do not follow verb_noun pattern. Names like 'iso_amt', 'nso', 'rsu', 'qsbs' are acronyms or abbreviations that do not convey action. Better names: 'calculate_iso_amt', 'analyze_nso_decision', 'evaluate_rsu_vesting', 'check_qsbs_qualification'. 90% of A+ tools start with action verbs.
No evidence of error handling patterns. Cannot verify that tools return recovery guidance (e.g., 'Invalid input. Try...'), categorize errors as retryable or user-fixable, or provide actionable error messages with invalid values and constraints.
Tool composition and chaining unclear. No evidence that tools return the IDs and references needed by downstream tools, or that tools are idempotent and safe for LLM retry patterns.