strale_methodology
Explains Strale's dual-profile scoring model for AI agent trust and quality, covering code quality, operational dependability, test infrastructure, audit trails, and transparent scoring methodology.
Instructions
Get Strale's quality and trust methodology. Explains the dual-profile scoring model: Quality Profile (code quality, 4 factors) and Reliability Profile (operational dependability, 4 factors weighted by capability type), combined via a published 5×5 matrix into the SQS confidence score. Covers execution guidance, test infrastructure (~1340 test suites with tiered scheduling), provenance tracking, audit trails, badge system, and honest disclosure of current limitations.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Implementation Reference
- packages/mcp-server/src/tools.ts:847-870 (handler)The tool 'strale_methodology' is registered directly in 'packages/mcp-server/src/tools.ts'. Its handler is defined inline as an asynchronous function returning a hardcoded string describing the methodology.
async () => { const methodologyText = `STRALE QUALITY & TRUST METHODOLOGY =================================== WHAT STRALE IS Strale is trust and quality infrastructure for AI agents. Agents call capabilities (atomic data operations) and solutions (multi-step workflows) via a unified API. Every execution is independently tested, scored, and auditable. SQS — STRALE QUALITY SCORE The SQS is a combined confidence score (0-100) derived from two independent profiles: - Quality Profile (QP): How well-built is Strale's code? (code correctness, schema compliance, error handling, edge cases) - Reliability Profile (RP): How dependable is the service right now? (availability, success rate, upstream health, latency) The two profiles combine via a published matrix into the headline SQS score. QUALITY PROFILE (QP) Measures code and methodology quality. Stable over time — only changes when code changes. Four factors: Correctness (50%) — Does it return accurate data for known inputs? Schema Compliance (31%) — Does the response match the declared format? Error Handling (13%) — Are errors caught and reported cleanly? Edge Cases (6%) — Does it handle unusual inputs gracefully? Upstream service failures are EXCLUDED from the Quality Profile. Grade scale: A (>=90), B (>=75), C (>=50), D (>=25), F (<25) Label format: "Code quality: [Grade]" (DEC-20260315-J) - packages/mcp-server/src/tools.ts:839-847 (registration)The 'strale_methodology' tool is registered using server.registerTool within the MCP server tool initialization logic in 'packages/mcp-server/src/tools.ts'.
// Meta-tool: strale_methodology (no API key required) server.registerTool( "strale_methodology", { description: `Get Strale's quality and trust methodology. Explains the dual-profile scoring model: Quality Profile (code quality, 4 factors) and Reliability Profile (operational dependability, 4 factors weighted by capability type), combined via a published 5×5 matrix into the SQS confidence score. Covers execution guidance, test infrastructure (~${capabilities.length * 5} test suites with tiered scheduling), provenance tracking, audit trails, badge system, and honest disclosure of current limitations.`, inputSchema: z.object({}), }, async () => {