MCP & Developer Skills

CONCEPTUAL DOCUMENTATION

Runtime is in development. These guides explain design intent; they are not a released API or installation reference. Read the release policy →

Planned, not installable

ZERNOVA AI is a planned product direction covering Model Context Protocol interfaces and Developer Skills. No MCP server URL, tool schema, skill bundle, install command, or integration with a named assistant is released here.

Give tools useful, bounded context

The intended direction is to help developer tools understand integration concepts, relevant contracts, and future verified examples. Context should identify its version and source, and distinguish instructions from untrusted content.

An assistant’s suggestion is not evidence that a flow is valid. Future tooling should connect changes to validation and reproducible tests.

Separate reading from acting

A documentation lookup and a production deployment have different consequences. Tool design should make permissions and side effects clear, use the smallest useful scope, and keep sensitive material out of returned context.

Action class	Design consideration
Read documentation	Version, source, and relevance
Propose a change	Reviewable diff and stated assumptions
Validate locally	Isolated environment and bounded resources
Change a live system	Explicit authority, evidence, rollback, and appropriate review

Make skills reviewable

A future Developer Skill should describe its purpose, prerequisites, allowed actions, verification steps, and stopping conditions. It should not embed secrets or silently grant production access.

Publication requirements

Released tooling will need a versioned contract, documented clients, permission model, tests for failure and malicious input, and a clear installation source. Until then, this page records design intent only.