Loading ToolBench
Agent-ready research server for strategy due diligence. Analyzes historical trade exports, ranks strategies, generates reports, and surfaces overfitting/risk blockers before paper trading.
Nexural Automation provides 8 tools with complete input schemas and descriptions, but quality is uneven. Tool names follow verb_noun convention (list_, analyze_, compare_, generate_, run_, estimate_, scaffold_). Descriptions range from 50 - 200 chars, meeting baseline length. However, parameter descriptions are sparse or generic ('Path to the trade export CSV file' lacks format/constraint details). Output schemas are not documented, LLMs cannot predict return structure. Error handling is absent; no guidance on retryability, user-fixable vs fatal errors, or recovery paths. Risk annotations (READ_ONLY, WRITE) are present but not formalized as toolAnnotations. No evidence of idempotency guarantees, pagination support, or chaining-ID inclusion in responses.
Run full strategy due diligence on a supported trade-export CSV. Returns a decision gate, summary metrics, overfitting checks, Monte Carlo risk envelope, walk-forward validation, and a prioritized improvement plan.
Rank 2-10 strategy exports by composite institutional metrics.
Estimate futures round-turn commission and slippage for a strategy.
Generate a local HTML research report for a supported strategy CSV.
List supported imports, workflows, and safety guardrails.
Run the institutional 10-check validation gauntlet on a trade CSV.
Output schemas not documented. LLMs cannot predict return structure, field names, or types. Downstream tool chaining is impossible without trial-and-error.
Parameter descriptions lack format/constraint details. E.g., 'Path to the trade export CSV file' does not specify file format, required columns, or encoding. 'Number of trials' lacks min/max bounds.
No error handling guidance. Tools do not document retryability, user-fixable errors, or recovery paths. E.g., if CSV is malformed, what should the LLM do next?
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | 2026-07-28+ | v2 |
Create a bridge SDK scaffold with contract and starter connector.
Create a strategy SDK scaffold with docs, metadata, and starter source.
Risk annotations (READ_ONLY, WRITE) are informal comments, not formalized as MCP toolAnnotations (readOnlyHint, destructiveHint). Agents cannot programmatically detect side effects.
No pagination or result-limiting strategy documented. Tools like compare_strategy_csvs accept unbounded arrays; no guidance on max items or chunking.