provider-capability-envelope

Resolve provider runtime flags to prevent dashboard drift.

Updated Mar 13, 2026
One-click install
npx skills add https://github.com/JDL440/nfl-eval --skill provider-capability-envelope
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: provider-capability-envelope
Source: https://github.com/JDL440/nfl-eval/tree/main/.squad/skills/provider-capability-envelope
Command: npx skills add https://github.com/JDL440/nfl-eval --skill provider-capability-envelope

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Prevent dashboard and trace-contract drift when a provider has both a high-level mode switch and lower-level runtime gates.

Core Features & Use Cases

  • Inspect the server/provider wiring to find the explicit mode input and any legacy compatibility flags.
  • Resolve the provider’s runtime flags first and build prompt/tool-policy text from those resolved flags rather than raw config booleans.
  • Preview the resolved route before building the tool contract so the runtime can decide capability from the executing provider rather than the caller’s guess.
  • Keep provider adapters honest about structured-output differences when the tool loop depends on machine-readable turns.
  • Add regression tests where explicit mode contradicts legacy booleans so the modern contract wins, and tests for fail-closed runtime conditions.

Quick Start

Review the server/provider wiring to identify how tool access mode is configured and ensure the runtime flags are resolved before prompts are generated.

Frequently Asked Questions about provider-capability-envelope

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent dashboard drift when provider config contradicts runtime flags?

Dashboard drift is prevented by resolving provider runtime flags first and building prompt or tool-policy text from those resolved flags rather than raw config booleans.

What is a provider capability envelope in contract validation?

A provider capability envelope aligns configured provider access with actual runtime capability by persisting both configured intent and resolved capability in the request envelope.

Why does my dashboard tool contract not match the actual provider mode?

This mismatch occurs when tool contracts are built from raw config booleans instead of the provider's resolved runtime flags, bypassing the actual executing provider's capability.

How do I add regression tests for explicit mode and legacy runtime flag conflicts?

Add regression tests where explicit mode contradicts legacy compatibility booleans to ensure the modern contract wins, and include tests for fail-closed runtime conditions.

Do I need to resolve runtime flags before generating prompts for provider tools?

Yes, you must resolve the provider's runtime flags first to preview the resolved route before building the tool contract, allowing the runtime to decide capability correctly.

Can I use structured-output tool loops with providers that have legacy compatibility flags?

Yes, but you must keep provider adapters honest about structured-output differences when the tool loop depends on machine-readable turns and legacy flags are present.