sdd-apply

Implements tasks from spec-driven change proposals by writing code following specs and design.

Updated Apr 4, 2026
One-click install
npx skills add https://github.com/juanjo-zurich/juarvis-v4 --skill sdd-apply-juanjo-zurich
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sdd-apply
Source: https://github.com/juanjo-zurich/juarvis-v4/tree/main/plugins/sdd/skills/sdd-apply
Command: npx skills add https://github.com/juanjo-zurich/juarvis-v4 --skill sdd-apply-juanjo-zurich

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? In Spec-Driven Development (SDD) workflows, translating approved specs, designs, and task lists into actual working code is error-prone: agents skip reading specs, deviate from design decisions, or lose track of task completion. This Skill acts as the implementation sub-agent that strictly follows the specs and design artifacts, marks tasks complete as it goes, and persists progress so downstream verification steps work. ## Core Features & Use Cases - Spec-Guided Implementation: Reads proposal, spec, design, and tasks artifacts (from Engram memory, OpenSpec files, or hybrid storage) before writing any code, treating spec scenarios as acceptance criteria. - TDD and Standard Modes: Auto-detects TDD configuration and enforces the RED-GREEN-REFACTOR cycle, or falls back to a standard write-then-verify workflow. - Progress Persistence: Updates tasks.md with completion marks and saves progress artifacts via mem_save/mem_update so the sdd-verify step can find implementation state. - Use Case: An orchestrator assigns "Phase 1, tasks 1.1-1.3" of an auth change; this Skill loads the skill registry, reads the JWT middleware spec, writes failing tests, implements the middleware, marks tasks complete, and returns a structured progress report. ## Quick Start Ask the agent to implement tasks 1.1 through 1.3 from the current change's tasks.md following the specs and design, then report progress.

Frequently Asked Questions about sdd-apply

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

FAQPage Schema
How do I implement tasks from an OpenSpec change proposal?▼

Read the proposal, spec, design, and tasks artifacts first, then implement each assigned task following the design decisions. Mark each task complete with [x] in tasks.md as you go, and return a structured summary of files changed and remaining work.

How does TDD mode work in a spec-driven workflow?▼

TDD mode is detected from openspec/config.yaml rules.apply.tdd, installed TDD skills, or existing test patterns. Each task then follows RED (write a failing test from spec scenarios), GREEN (write minimal passing code), and REFACTOR (clean up while tests stay green).

What is the difference between engram, openspec, and hybrid artifact modes?▼

Engram mode stores artifacts in persistent memory via mem_search and mem_get_observation, openspec mode uses filesystem files like tasks.md, and hybrid mode does both. Mode none returns progress only without updating project artifacts.

Why does mem_search return incomplete spec content?▼

mem_search returns only 300-character previews, not full content. You must call mem_get_observation with the observation ID for every artifact to retrieve the complete proposal, spec, design, and tasks before implementing.

What happens if implementation progress is not persisted?▼

If you skip saving progress via mem_update and mem_save, the downstream sdd-verify step cannot find the implementation state and the SDD pipeline breaks. Persisting progress after marking tasks complete is mandatory.