Skip to main content
Prompts you can paste straight into your AI assistant, followed by worked examples showing what actually comes back. For every tool and what it does, see the Tools Reference.

​
10 prompts to try

One prompt per area, covering everything Feedspace can do. Paste any of these in and adjust the names and filters to match your own workspace.
New here? Start with List my Feedspace workspaces. If it returns your workspaces, you are connected and everything else will work.
Prompts marked View only work in every AI tool. Ones marked Makes changes need a tool that is allowed to create and edit things.Most tools can do both. Some managed setups restrict what a connector may change, so if reading works but creating a widget does nothing, check with whoever administers your AI tool before assuming it is a Feedspace problem. You can always make the change in the Feedspace dashboard.
Filtering has no date range. You can narrow reviews by rating, sentiment, type, form, label, keyword, and whether the reviewer gave a name, company, or photo, but not by “the last 3 months”. Sort or trim by date after the assistant has fetched them.

​
Worked examples

These show the full round trip: what you ask, what the assistant does, and what you get back.
The replies below are illustrative. Your own counts, names, and links will differ, and each assistant words things its own way. What is accurate is the shape of the answer and the tools involved.
Makes changes. Creates a widget.You: “From my 5-star text reviews, pick the 5 best for a homepage carousel. One line on why each works. Then create the widget and give me the embed code.”What the assistant does:
  1. filter_reviews with starRatings: [5], reviewTypes: ["text"]
  2. Reads them and picks the 5 most persuasive
  3. create_widget with type: 4 (Carousel) and the feed_identifier of each
  4. get_widget_share for the embed snippet
What you get back:
Picked 5 from your 5-star text reviews:
  • Sarah Chen names a result, “cut our onboarding time in half”
  • Miguel Torres answers the usual objection about setup effort
  • Priya Nair credits the support team by name, so it reads as genuine
  • Tom Alder mentions switching from a competitor
  • Ana Ruiz carries authority, “sets a new standard”
Creating the widget now. “Homepage Carousel” is live in My First Workspace with your 5 hand-picked reviews. Here’s the embed code:
Paste this anywhere on your homepage HTML and the carousel will render there.
Say how many you want. “Pick the 5 best” keeps the answer short. Without a number, the assistant pages through everything to be thorough and can return dozens.Ask for the embed code in the same prompt. Creating a widget and fetching its snippet are separate tools, so without those words the assistant creates the widget and then asks whether you want the code, costing a round trip. Same for get_page_share and get_form_share.
Results are paged, 15 at a time. filter_reviews ignores perPage, so the assistant works through pages and may stop before it has everything. It will usually say more remain. Reply “keep going” to pull the rest, or narrow the filter.
View only. Works in every AI tool.You: “How are my reviews doing? Give me the key numbers, then two things working and two that need attention.”What the assistant does:
  1. get_review_summary for the total count
  2. filter_reviews with starRatings: [1, 2] to see what is going wrong
  3. list_forms for views, submissions, and conversion rate per form
  4. list_widgets for view counts and engagement rate per widget
  5. list_pages for Wall of Love view counts
What you get back:
Key numbers
  • 412 total reviews
  • At least 9 reviews with 1 to 2 stars found in a single search pass, likely more
  • 3 review forms, your best converting at 15.8%; the other two are under 4%
  • 1 Wall of Love page and 4 widgets live
Two things working
  1. Your 5-star reviews are genuinely strong. They are detailed, name real people, titles, and companies, and cite specific outcomes (one has account setup dropping from 7 months to 2 days). That is exactly what converts on a homepage.
  2. You are turning reviews into assets. Between the Wall of Love and the widgets, you have several live, embeddable showcases. Most workspaces just let reviews sit uncollected.
Two things needing attention
  1. Two forms get traffic but almost no submissions. Both sit under 4%, so the form is likely too long or the ask is unclear. Your best form shows what good looks like.
  2. Two widgets have no views. They are probably not embedded yet, so the reviews behind them are not doing any work for you.
Notice “found in a single search pass, likely more”. The assistant reports what it actually looked at rather than implying it read everything, because results come back 15 at a time. Ask “check the rest” if you need a complete count.
Makes changes. Creates a label and tags reviews.You: “From my 5-star reviews, tag the 10 best for our homepage with a ‘Homepage’ label. Skip anything too short or vague, and tell me how many you skipped.”What the assistant does:
  1. create_label with the name “Homepage” and a hex colour
  2. filter_reviews with starRatings: [5]
  3. Reads them and picks the 10 worth featuring
  4. batch_update_reviews with action: "assign_labels" and those 10
What you get back:
Done, 10 reviews tagged “Homepage”:
  • Sarah Chen: speed and support, calls it “a new standard”
  • Miguel Torres: names the problem it replaced
  • Priya Nair: praises patient, helpful support
  • Tom Alder: time and cost saved for a small team
  • Ana Ruiz: a specific feature with a measurable result
  • Leo Fontaine: concrete numbers rather than adjectives
  • Mina Han: names the support rep who helped her
  • Nordic Supply: team praise tied to business impact
  • Hassan A.: specific outcome, and mentions the agent by name
  • Sam Okafor: answers common buyer objections directly
Skipped 6, all too short or vague to carry weight on a homepage:
  • Three one-word reviews such as “Great!” or a thumbs-up emoji
  • One duplicate of a review already in the list
  • One that hedges rather than endorses
  • One about a discount code rather than the product
One caveat: I have only read your 5-star reviews from the first couple of pages, not the whole set, so there may be more strong candidates. Want me to keep scanning?
Makes changes. Creates a page and automation rules.You: “Build a Wall of Love called ‘Customer Love’ from the 10 reviews that best show real results, not just praise. Keep it updated as new 5-star ones arrive, and give me the public link.”What the assistant does:
  1. filter_reviews to pull your highest-rated reviews
  2. Reads them and picks the 10 describing an outcome
  3. create_page with the name and those reviews
  4. update_page_automation with review_ratings: [5]
  5. get_page_share for the public link
What you get back:
Done. “Customer Love” is live in My First Workspace. Public link: https://www.feedspace.io/your-workspace/customer-love The 10 reviews, chosen for concrete, measurable outcomes rather than general praise:
  • Sarah Chen: account setup went from 6 weeks to 2 days
  • Miguel Torres: cut onboarding admin roughly in half
  • Sam Okafor: lower transaction fees, with figures
  • Priya Nair: fast setup, no fees, savings they can point to
  • Tom Alder: solved a pricing and minimum-balance problem
  • Ana Ruiz: streamlined payments across several accounts
  • Mina Han: real-time accounting sync instead of overnight updates
  • Leo Fontaine and Ines Roy: replaced their invoicing and tracking process outright
  • Hassan A.: support resolved a failed transaction with clear, concrete steps
  • Nordic Supply: card and account setup in minutes rather than days
Auto-update is on: any new 5-star review will be added to this page automatically.
Automation cannot judge quality. The 10 reviews above were hand-picked for showing a measurable result. Automation only filters on rating, type, labels, form, keywords, and reviewer details, so future additions are held to “5 stars” and nothing more. A short “Great!” will join the page.If the bar matters, tag the good ones with a label and set the automation to that label instead of to the star rating.
Makes changes. Creates a form.You: “Create a review form called ‘Customer Feedback’ that accepts written and video reviews, asks for name, email, and company, and give me the link to share.”What the assistant does:
  1. create_review_collection_form with the name, the review types, and the reviewer fields
  2. get_form_share for the link
What you get back:
Done. Your “Customer Feedback” form is live in My First Workspace.
  • Accepts: written and video reviews
  • Asks for: name, email, company (all required)
Shareable link: https://www.feedspace.io/your-workspace/customer-feedback
Fields you ask for come back required. The reply above marks name, email, and company as required, which will cost you submissions from people who do not want to give all three. Say “make email optional” in the prompt, or ask for it afterwards.
Name the review types you do not want. Written, video, and audio are all enabled by default, so listing the ones you do want will not necessarily switch the others off. Say “no audio” if you mean it.Ask for the QR code if you want one. get_form_share returns a QR alongside the link, but the assistant usually shows only the link.
Makes changes. Adds a review.You: “I got this by email and want to use it on our site. Sarah Chen, Head of Design at Acme, said ‘The onboarding was the smoothest I’ve used.’ It was 12 March. Add it and put it on my Homepage Testimonials widget.”What the assistant does:
  1. import_text_review with the reviewer name, role, company, review text, rating, and date
  2. update_widget to add the new review to the widget you named
What you get back:
Review imported (Sarah Chen, Head of Design at Acme, 5★, dated March 12, 2026). Now adding it to your Homepage Testimonials widget. Done. Sarah Chen’s review is now imported and added to the Homepage Testimonials widget, which now shows 9 reviews.
Give the year for anything older than this one. “12 March” was read as March 2026, the current year. An older testimonial needs the full date, or it will be filed under today’s year and sit at the top of a newest-first list.Say the rating if you care. No rating was mentioned, so it was saved as 5 stars, the same default the web app’s import form uses. Half stars such as 4.5 are accepted.

​
Tips for best results

  • Name your workspace if you have more than one, for example “in my Acme workspace”. Otherwise the assistant has to ask.
  • Say the review type when filtering: “text reviews”, “video reviews”, “audio reviews”, or “social reviews”.
  • Ask for the link when you want something shareable, such as a form link, page URL, or embed code. The assistant will not always fetch it unprompted.
  • Chain steps in one prompt. “Find my best reviews and build a Wall of Love from them” works in a single request.
  • Ask it to check first before a bulk change, for example “show me which reviews that would affect before you apply the label”.