Live Bitcoin market regime for agents: three sensors read pre-price data — fee curves, funding velocity, whale flows, absence — every 30s and reject almost everything. When they converge, a directional call is emitted with evidence. The data can't be reconstructed from history. Public, agent-auditable track record. MCP + REST + x402. Free tier, no card.
BRS Signals MCP exposes 7 tools with inconsistent definition quality. Tool names follow verb_noun convention (brs_market_state, brs_decision_context) but lack clarity on tier/access model in the name itself. Descriptions are present and domain-specific (10 - 200 chars), but parameter schemas are largely absent from visible code. Only brs_raw_stream shows explicit input schema (stream enum). Most tools lack documented output schemas, parameter descriptions, and error recovery guidance. The server implements per-client API key pass-through and x402 payment challenges (novel auth patterns), but these are not reflected in tool definitions or parameter constraints. No tool annotations (readOnlyHint, destructiveHint) despite all tools being READ_ONLY. Telemetry logging is implemented but not exposed as a tool parameter or configuration option.
Free tier tool that returns public, keyless audit track record of all signals with evidence for agent auditability
Pro tier metered tool that returns per-call bias context and decision evidence; callable keyless with x402 payment challenge for one-off paid calls
Pro tier tool that returns historical bias data and decision context over time
Free tier tool that returns live Bitcoin market structure readings combining fee curves, funding velocity, and whale flows with system health status
Pro tier tool that returns raw streaming data from fee curves, funding velocity, stablecoin flows, and gamma sensors
Free tier tool that returns the public rejection funnel showing gate-by-gate filtering of market readings
No input schemas visible for 6 of 7 tools. Only brs_raw_stream declares an explicit input schema (stream enum). Remaining tools lack documented parameters, types, and constraints, forcing LLMs to guess at valid inputs.
No output schemas documented. Tools return API responses (structure, confidence, bias, track-record, funnel, health, streams) but LLMs cannot plan downstream calls or extract required fields without knowing the response structure.
Tool names do not distinguish tier/access model. 'brs_decision_context' and 'brs_raw_stream' are Pro tier (metered, x402 payment), while others are Free. LLMs cannot infer cost or auth requirements from names alone. Recommend suffixes: brs_decision_context_pro, or explicit tier in description.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | 2026-07-28+ | v2 |
Free tier tool that returns system health status including SLO metrics and operational counters
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite all tools being READ_ONLY. Annotations enable agents to reason about safety and retry logic. All 7 tools should declare readOnlyHint=true.
x402 payment challenge and per-client API key pass-through are implemented server-side but not exposed as tool parameters or documented in parameter descriptions. LLMs cannot understand how to provide keys or handle payment challenges without explicit parameter documentation.