Your First Intent in Pathmode
If you're reading this, you probably understand the philosophy of Intent Engineering. But how do you actually do it in Pathmode?
Here is a step-by-step guide to creating your very first Agent-Ready IntentSpec using the Pathmode UI.
Capture the Evidence
Intent Engineering starts with a problem, not a solution. Before you write a single line of spec, you need proof.
- Navigate to the Evidence tab in your workspace.
- Click Add Evidence.
- Choose a type (e.g.,
friction,quote,metric). - Paste in real feedback from a user. For example: "I tried to pay but the screen just hung there for 10 seconds. I had no idea if my card was charged."
- Save the evidence item.
Open the Intent Editor
Now that we have proof of friction, let's solve it.
- Navigate to the Specs tab.
- Click New Intent.
- You're now in the Intent Editor. This isn't a blank Google Doc; it's a structured interface designed to build a complete
IntentSpecobject.
Define the Core Problem
The top of the editor asks for the fundamental "what" and "why."
- Title: Keep it punchy. e.g., "Fix checkout payment timeout"
- Objective: Explain the problem and why it matters. e.g., "Reduce cart abandonment caused by payment step timeouts exceeding 3 seconds"
- Severity: Set it to
critical(since it's a payment issue).
Map the Outcomes
Scroll down to the Outcomes section. Instead of writing tasks ("add a spinner"), write the observable state changes.
Click Add Outcome for each:
- "Payment completes in under 3s (p95)"
- "Users see real-time status during processing"
- "Failed payments show actionable error with retry"
Anchor the Evidence
Remember that evidence we captured in Step 1? Time to use it.
- In the right-hand panel, open the Evidence tab.
- Search for the payment friction you added earlier.
- Click the Anchor icon next to the evidence, then select the Objective field in your spec.
- You have now explicitly linked the "Why" to real user pain. The "Solutioneering" risk is neutralized.
Set Guardrails
What could the AI agent do wrong when fixing this? Let's constrain it.
- Scroll to Constraints.
- Add: "No double-charge on retry"
- Add: "Must support iOS Safari 15+"
Define Edge Cases
Happy paths are easy. Let's cover the failures.
- Go to Edge Cases.
- Add a Scenario: "Network timeout during payment"
- Add Expected Behavior: "Show retry button, no double-charge"
Define Verification
How will the agent prove it succeeded?
- Go to Verification.
- Add an End-to-End Test requirement: "checkout-payment-retry.spec.ts"
Export to Agent
You're done! Your spec is now formally structured, validated against constraints, and backed by evidence.
Click Export (or use the Pathmode MCP Server) to inject this IntentSpec directly into Claude Code or Cursor. The AI now knows exactly what to do, why to do it, and what it cannot do.