project-architect

Turn a project idea into implementation-ready documentation by running an interactive discovery pipeline that generates specification, implementation, and task files. This structured planning approach outputs a single-shot agent prompt with explicit file-level targets and acceptance criteria to prevent ambiguity-driven rework.

Plan a tech stack and select design patterns through a documentation-first consultation workflow. The pipeline reads reference guides to evaluate constraints and architecture direction, outputting a blueprint that defines your stack and pattern choices before generating implementation tasks.

A documentation-first planning pipeline works by guiding structured requirements gathering for scope, audience, and constraints, then producing a reviewable sequence of interconnected documents like SPECIFICATION.md, IMPLEMENTATION.md, and TASKS.md before synthesizing a final autonomous execution prompt.

You can generate only a specification document without the full blueprint. The pipeline supports partial generation modes, allowing you to request spec-only, prompt-only, or stack and pattern consultation outputs rather than forcing a complete end-to-end documentation run.

You need YAML-defined skill metadata to initiate the sequential planning pipeline. This metadata structures the reference reads for elicitation, tech stack selection, patterns, specification, implementation, tasks, branding, and final prompt synthesis required to produce architecture blueprints.

Use stepwise review and approval flows for task breakdown to catch ambiguity early in planning. By reviewing interconnected specification and implementation documents sequentially before generating the final agent prompt, you prevent iterative rework during the actual coding phase.