Starter prompts
4 ways to start with AgentSearch.
Audit
→ Can AI do it
▸ Preview prompt
Audit our site for agentic task completion — can a browsing agent book, buy, register, or subscribe without human help?
WebMCP
→ Implement
▸ Preview prompt
Add WebMCP declarative + imperative patterns to our booking flow so AI agents complete bookings reliably.
Friction
→ Find the breaks
▸ Preview prompt
Identify every friction point that blocks AI agents on our checkout — CAPTCHAs, hidden forms, JS-only state, modal traps.
Metrics
→ Track success
▸ Preview prompt
Set up agent traffic identification and task completion tracking — what events to log, what dashboards to build.
What it does
Tasks AgentSearch ships every week.
WebMCP
- Declarative WebMCP patterns
- Imperative action exposure
- Schema + intent metadata
- Agent-readable forms
Measurement
- Task completion rate
- Agent traffic identification
- Friction point audit
- Browsing agent compat
Worked sample
A real AgentSearch chat.
Pairs well with