Intent: Why
Capture the strategic motivation through structured envisioning. Questions surface customer pain points, goals, and business context before any code is written.
DevSquad Copilot
Ship at the pace of innovation with agents that keep delivery grounded in specs, traceable decisions, and production-ready quality.
Delivery as a Feedback Loop
Section titled “Delivery as a Feedback Loop”Shipping at the pace of innovation means treating delivery as a loop, not a pipeline. The framework organizes work in four layers. Specs are sliced thin and revised as you learn; each layer produces artifacts that reduce rework downstream and stay amendable when implementation reveals new understanding.
Specification: What
Feature and migration specs define scope, acceptance criteria, and constraints for the current slice. Written for stakeholders, these artifacts are the contract downstream work is measured against, and amendable when reality shifts.
Architecture Decisions: How
ADRs record how and why specific technical approaches were selected. Evaluated against ranked priorities, not generic pros/cons, so decisions stay traceable.
Delivery: When
Every increment is traceable and reviewable. When implementation reveals that a spec or ADR no longer matches reality, the framework amends the affected section and continues, instead of silently deviating.
Pace You Can Keep
Section titled “Pace You Can Keep”Four concrete mechanisms separate DevSquad from “describe the whole system, then let the AI code it” tools. Each one keeps delivery reviewable and amendable slice by slice, so velocity holds up past the first few iterations.
Thin slices, amendable mid-flight
Specs are written at the slice level and amended when implementation reveals a model shift. The amendment seam is first-class, not a retrofit.
Decisions as ADRs, not code comments
Every non-trivial technical choice produces an ADR evaluated against ranked priorities. Reviewers and future agents read the reasoning, not just the code.
Impact-scaled ceremony
Low-impact changes move fast. Medium and high changes trigger the checks they need: plan review, ADR, explicit approval. Ceremony matches risk.
Comprehension checkpoints
After implementation, the framework checks that the developer understands what was generated and why. The loop optimizes for comprehension, not throughput.
Who This Is For
Section titled “Who This Is For”A good fit if you...
- Ship production code that other people depend on.
- Want AI to accelerate delivery without dropping specs, ADRs, or review.
- Work in a team where traceability and onboarding matter.
- Are comfortable reviewing AI output at every impact-classified step.
Not a fit if you...
- Want unsupervised end-to-end code generation from a prompt.
- Are prototyping throwaway code where ceremony is pure overhead.
- Prefer to treat AI as a code autocomplete without structured handoffs.
- Expect the tool to pick architecture for you without developer input.
Start Here
Section titled “Start Here”Pick the entry point that matches your role. Each path includes the essentials for that audience.
Developers
Install the plugin and run your first guided workflow.
Architects
Leadership
Evaluate how guardrails keep delivery predictable without slowing it down.
Platform Teams
Extend DevSquad with your own instructions, skills, agents, and hooks.
Find this framework useful?