Open a chat with your AI agent inside the target repository and paste the command for your agent:
@workspace Follow the instructions at https://raw.githubusercontent.com/markhazleton/devspark/main/quickstart/devspark_quickstart_copilot.md
Follow the instructions at https://raw.githubusercontent.com/markhazleton/devspark/main/quickstart/devspark_quickstart_claudecode.md
Follow the instructions at https://raw.githubusercontent.com/markhazleton/devspark/main/quickstart/devspark_quickstart_cursor.md
Follow the instructions at https://raw.githubusercontent.com/markhazleton/devspark/main/quickstart/devspark_quickstart_codex.md
For Codex-specific workflow guidance, see DevSpark and Codex.
Follow the instructions at https://raw.githubusercontent.com/markhazleton/devspark/main/quickstart/devspark_quickstart_antigravity.md
For Antigravity-specific workflow guidance, see DevSpark and Antigravity.
Follow the instructions at https://raw.githubusercontent.com/markhazleton/devspark/main/quickstart/devspark_quickstart_generic.md
The agent first asks only the install-critical questions, checks for existing DevSpark or legacy layouts, and only asks for project name, tech stack, and core principles when a constitution still needs to be created. No CLI required.
Need terminal-driven setup or execution? See Other Ways to Get Started for CLI installation options, then use the runtime guidance in Harness Engineering.
Once bootstrapped, run these slash commands in your AI agent's chat.
/devspark.constitution Security-first. TDD required. All public APIs must have documentation.
Describe what you want to build and why -- no tech stack yet. Keep it product-focused.
/devspark.specify Build a photo album organizer. Albums grouped by date, drag-and-drop reordering, tile-based photo previews.
/devspark.specify is route-aware. It recommends a one-off fix, quick spec, or full spec path, explains why, and asks you to confirm before it creates artifacts.
Anti-pattern:
/devspark.specify Build a React app with Redux and PostgreSQL for photo management-- this locks you into a solution before the problem is fully understood.
Ask about user needs and constraints, not implementation details.
/devspark.clarify Focus on security and performance requirements.
Anti-pattern:
/devspark.clarify Should we use WebSockets or SSE?-- save technology choices for the plan phase.
Now provide your tech stack. The plan translates product requirements into architecture.
/devspark.plan Use Vite with vanilla HTML/CSS/JS. Images stored locally, metadata in SQLite.
/devspark.tasks
Optionally validate first with /devspark.analyze or /devspark.critic, then:
/devspark.implement
After implementation completes (spec status becomes Complete), draft the PR and review it:
/devspark.create-pr
/devspark.pr-review
If you push more commits after review feedback or a rebase, refresh the description with /devspark.update-pr before re-reviewing.
The review checks that the spec is Complete and all tasks are done before recommending approval. Merge the PR after approval.
At the end of the sprint, archive completed specs and generate release notes:
/devspark.release
See Implementation Lifecycle for the full spec status lifecycle and sprint cadence.
- Upgrade Guide -- keep DevSpark current
- DevSpark and Codex -- best practices for using Codex with DevSpark
- Implementation Lifecycle -- full workflow overview
- Harness Engineering -- optional CLI runtime for declarative workflows
- Harness Strict Template -- delivery-integrity defaults for hands-off runs
- Constitution Guide -- writing effective project principles
- FAQ -- common questions answered
- Command Reference -- all 28 commands
For delivery-integrity focused execution, use the strict harness template and run with an explicit write-capable adapter:
devspark harness run sample.harness.yaml --hands-off --adapter claude_code
If the latest run is not create-pr ready, inspect decision-packet.json and no-change-explainer.md before continuing.