Who writes it
Should It Be a Prompt? is written by Mahmoud Halat, founder of Space & Story, a web and AI studio in Toronto. A lot of the work with clients is deciding what to build, and a good share of that is talking a request down from an app to a skill, or up from a prompt to something with a connector and an approval step. This site is that conversation written down, so a team can have it before the kickoff call. I also won the Grand Prize in Lovable's Shipped season 1, which is relevant to the app builders comparison and disclosed there.
What we own, and how we make money
We say this on every page where it matters, and here in full:
- Space & Story is my studio. It builds web and AI products for clients, and it's linked once, on the app builders page, as an option for when a project outgrows the builders.
- Skillspace (skillspace.sh) is a directory that rates public agent skills. It's ours; its site lists XY Space Inc. as the operator. It appears on the registry comparison, the MCP and skill vetting guide and in the pattern library, with its weaknesses stated alongside everyone else's. It isn't an app builder or coding agent, so it isn't scored in the reviews.
- We don't take affiliate commissions, sponsorships or payment for placement from any product we compare. No link on this site pays us when you click it or buy something.
The site itself makes no money directly. It exists because a clear answer to "should this be a prompt?" helps the people we'd like to work with, and because it keeps us honest about our own products.
How the decision tool works
The decision tool applies a fixed set of rules to eight answers. There's no model call, no scoring weights and no data collection: your answers stay in your browser, and the only thing that leaves it is the page request itself, counted by our analytics. Every rule is published in a table on the tool's page. The pattern library runs 38 tasks through the same rules, and a test in the site's code fails the build if any pattern's stated verdict differs from what the tool returns for its answers.
The effort and upkeep estimates on each rung ("an afternoon", "days to weeks") are our judgement from building these things, not measurements, and we label them that way where they appear.
How we research comparisons
Every price, plan limit and feature claim in a comparison comes from the vendor's own pricing page, documentation or help centre, opened on the date shown on the page, and linked next to the claim. Where an official page didn't show a figure, we say where the figure came from instead, or leave it out. Where a vendor's own pages disagree with each other, we say so. Directory sizes are the numbers each directory shows about itself, except the official MCP Registry, which we counted ourselves through its public API.
We compare published specs and documentation. Where I've used a product myself, the page says so; where I haven't, it doesn't pretend otherwise.
Reviews are scored on a six-part rubric (speed, depth, backend, exit, pricing predictability and team readiness) that's published in full, with our update cadence and corrections policy, on the methodology page. Every review, head-to-head and table reads from one product dataset, so a price is the same wherever it appears. What changed and when is on the updates page.
What happened to the old site
Until October 2026 this domain ran an AI tool that judged startup ideas against the Moat Line. That product is retired. The essays behind it are kept here, lightly edited, and old links redirect to their new homes.
Corrections
Plans and prices in this market change monthly. If something here is out of date or wrong, tell me through mahmoudhalat.com and I'll fix it and update the date on the page.