How-to guides
Each guide here shows you how to do one thing in Intent Architect, in a short set of steps.
Getting Started
New to Intent Architect? Start here.
- Make your first change with the AIOpen your code, ask the AI for a change, then review and commit it without leaving the app.
- Use Claude Code, Cursor or Copilot with Intent ArchitectKeep the coding agent you already use, from your terminal, your editor or Intent Architect's chat.
- Ask the AI when you're stuckGet answers about Intent Architect and your own solution, linked to the docs they came from.
Working with the AI
Give the agent what it needs, and get more back from it.
- Give the AI the right context, fastPoint the AI at an element, a file, a screenshot or a work item instead of pasting walls of text.
- Connect an MCP server to your chatLet the agent read and update Azure DevOps, Jira, Slack and more, in the same chat as your code.
- Work across related codebasesGive the agent a frontend, a shared library or other code outside your solution to read or change.
- Reuse your best prompts with Work Item templatesWrite a prompt you keep retyping once, and start a fresh chat from it whenever you need it.
- Get visual reports on your system from the agentAsk about part of your system and get diagrams, tables and links to your model and code.
- Build your own dashboards and dev tools with the agentHave the agent build a dashboard of your pipelines, pull requests or anything else, with live data.
Configuring Your Workspace
Set up your workspace once so the agent can build, test and run your app.
Planning and Spec-Driven Development
Choose how much structure a change needs, from a quick prompt to a spec that stays true.
- Vibe coding, Plan Mode or Spec-Driven Development: when to use whichPrompt, plan or write a spec: pick the approach that fits the outcome you're after.
- Get it right the first time with Plan ModeCheck the data model, API and screens in the plan, and comment on it before any code is written.
- Turn existing code into living specsWrite down the rules your code already follows, back them with tests, and catch changes that break them.
- Use Spec Kit, Kiro, BMAD or OpenSpec with Intent ArchitectBring in the specs you have, run the framework's own commands, and see which code realizes each requirement.
Running Agents in Parallel
Keep several agents working at once, each in its own copy of your code.
- Run agents in parallelHand tasks to agents that each work in their own worktree while you carry on with yours.
- Stay on top of many agents at onceSee which agents are waiting on you, deal with them from the keyboard, and put away what's finished.
- Split a big task across child chatsHave the agent split a big task into chats that run side by side, then bring the results together.
Reviewing and Shipping
Review changes with confidence and get them merged.
- Review pull requests effectivelyLet AI review first, check the design on a diagram, and spend your time where a person is needed.
- Open pull requests that prove they workHave the agent prove its change with your tests, an AI review and your running app, then open the PR.
- Classify your codebase so reviews focus on what mattersTag your code by purpose and review importance, so you know which files in a large change to read first.
- Resolve merge conflicts with AIHand a merge, rebase or pull request update that stopped on conflicts to the agent to finish.