Loading ToolBench
A Model Context Protocol (MCP) server that provides funding-stress judgment, money-market evidence, capital-market transmission, and China macro research tools. Exposes the same board a human sees: current stress regime, forward odds of funding events, historical analogs, diagnostic status, and Time Machine replay.
Seiche provides 22 tools with consistent naming (verb_noun pattern: latest_article, funding_stress_now, etc.) and descriptions present for all tools. However, most tools lack input parameter descriptions, 10 of 22 tools have empty input schemas ({}), violating the pattern:tool-description requirement that every parameter needs a description. Tools 10 - 11 (money_market_context, world_markets_context) have parameters with types and descriptions, but output schemas are not documented anywhere in the source. Error handling and recovery guidance are absent. Descriptions average ~100 chars, which is acceptable but lack actionable context (e.g., 'when to use this instead of similar tools'). No tool declares permissions, audit trails, or security boundaries. The 'authenticated only' markers in descriptions are informal and not machine-parseable.
Append a signed entry to Agent Room chain (authenticated only, identity-bound)
Create a new Agent Room for private non-executable preview (authenticated only, identity-bound)
List verified entries in Agent Room chain (authenticated only, identity-bound)
Register a public key for Agent Room identity (authenticated only, identity-bound)
Verify complete chain integrity of Agent Room (authenticated only, identity-bound)
Query the desk for custom analysis and interpretation (authenticated only)
Cryptocurrency market stress indicators and historical record
10 of 22 tools have empty input schemas ({}), violating pattern:tool-description. LLMs cannot infer parameter meaning from names alone. Even no-parameter tools should document what they return.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields without knowing response structure. Violates pattern:tool and pattern:response-shaper.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | 2026-07-28+ | v2 |
Freshness and completeness status of input data sources
Authenticated desk brief with forward odds and detailed analysis (authenticated only)
Forward odds and probability forecast of funding stress events (authenticated only)
Current funding-stress regime assessment with probability and regime classification
Foreign exchange and materials market transmission evidence
Nearest historical analogs to current market conditions with precedent analysis
Institutional capital flows and positioning evidence with method versions withheld
Retrieve the latest published article from the Seiche editorial feed
USD money-market desk evidence with granular funding, repo, and liquidity data (cache-only, no collection trigger)
Oil market funding conditions and leverage positioning context
Current positioning book and leveraged positioning evidence (authenticated only)
Historical backtest record of the funding-stress model with track record validation
Time Machine replay of market conditions as of a specific date (authenticated only)
Trade execution safety and counterparty risk assessment
Cross-market transmission and capital-market evidence (cache-only, no collection trigger)
No error handling or recovery guidance. Tools lack actionable error messages (e.g., 'User not found. Try search_users() first.'). Violates pattern:recovery-guide.
Authentication and permission boundaries are informal ('authenticated only' in descriptions). No machine-parseable scope declarations (e.g., 'read:market-data', 'write:agent-room'). Violates pattern:scope-declaration.
Descriptions lack context for tool selection. No guidance on 'when to use this instead of similar tools' (e.g., funding_stress_now vs funding_stress_forecast). Violates pattern:tool-description completeness.