Is there a plan for a unified architecture guide for building harnesses with Spec Kit?

[waychow1986]

Hi Spec Kit maintainers,

We are experimenting with building a native Spec Kit based engineering harness using the existing building blocks:

• core SDD commands such as speckit.specify, speckit.plan, speckit.tasks, and speckit.implement
• presets for organization-specific templates, command overrides, terminology, and localization
• extensions for governance commands such as approval, verification, reporting, and handoff
• workflows for orchestrating commands, shell steps, and human gates
• deterministic scripts and CI for checks that should not rely only on agent prompts

The current docs explain these pieces separately, especially the references for presets, extensions, and workflows. What we could not find is a single architecture or methodology guide that explains how these pieces are intended to work together when building a governed engineering harness.

For example, a guide that covers questions like:

1. What should be handled by a preset versus an extension?
2. When should a workflow step be a command, a prompt, a shell step, or a gate?
3. What is the recommended pattern for deterministic quality gates and evidence collection?
4. How should approval boundaries before speckit.implement be modeled?
5. How should generated artifacts, workflow state, reports, and handoff notes relate to each other?
6. Are there recommended reference architectures for enterprise or team-level harnesses built on top of Spec Kit?

Is there already an official or recommended document for this that we missed? If not, is the Spec Kit team planning to provide a unified architecture guide or methodology for building harnesses with presets, extensions, workflows, scripts, and CI?

Thanks!

[mnriem]

We would love the community to work with us to create that! So if you have any suggestions or a PR to get the process started we would love to see that!