SpacetimeDB AX Audit: A Review for AI Agents

SpacetimeDB's homepage is written for AI agents. Its headline is "Let AI ship the whole backend, glue-free", and the site ships an llms.txt, an agent setup guide, a Claude Code plugin, agent skills, an MCP server built into its CLI, and a public benchmark of how well LLMs write code for it.

In our tests, a coding agent built a working, live-updating app on SpacetimeDB with no help from us. It used none of that tooling. And when we asked agents which backend to use for a web app, they picked Convex or Supabase.

SpacetimeDB is a database that is also your server, made by Clockwork Labs. We put it through our agent experience (AX) audit, following our AX audit method. The audit has four stages:

• Awareness: does the model know SpacetimeDB without searching?
• Discoverability: can an agent find SpacetimeDB when it's looking for a tool like it?
• Reputation: does the agent speak well of SpacetimeDB, and recommend it over other options?
• Integration: can the agent build a working proof of concept with SpacetimeDB?

How we tested

All testing used Claude Opus 5.5 (claude-opus-5-5) in Claude Code 2.1.285, on 7 October 2026. Every session started in a fresh, empty folder with no project files, memory, MCP servers, skills or plugins. Awareness questions ran with no tools at all, and discoverability and reputation questions with web search and web fetch only. The integration ran in a fresh Claude Code profile with no settings, plugins or history, with every command approved by us. We ran each prompt once, so treat these results as one developer's experience rather than a benchmark.

Scores at a glance

The model knows SpacetimeDB well, but as the game backend it was in 2025. It doesn't know the 2026 rebrand, the push into web apps or the agent tooling. Agents find it straight away for multiplayer games, but for web apps it comes second to Convex, and agents recommend Convex or Supabase over it. In the integration, the agent installed it, built a live page and hid private data without any help from us.

OKAwareness3 / 4Knows the games database well; doesn't know the rebrand, the web push or the agent toolingOKDiscoverability3 / 4Top pick for multiplayer games; second to Convex for web appsPOORReputation2 / 4Recommended for games; loses to Convex and Supabase for web appsGOODIntegration4 / 4Live app with private data, correct answers and no human stepsOKOverall3 / 4Agents build well on it, but don't choose it for the web apps it now targets

See our AX audit rubric for how we score each stage. The overall score is the average of the four.

Awareness: does the model know SpacetimeDB?

Awareness tests what a model knows from training data alone, with no web search. SpacetimeDB has been public since 2023 and reached 1.0 in March 2025, so we expected the model to know it. The open question was whether it knew what SpacetimeDB has become in 2026.

OK3 / 4Knows the games database well; doesn't know the rebrand or agent toolingFAILUnknown, or confused with something elsePOORRecognises the name, details vague or inventedOKAccurate description from memoryGOODAccurate, and recommended unprompted

We started with two questions that don't name SpacetimeDB. Each answer below comes from a fresh session with no tools.

Awareness · SpacetimeDB not namedWhat would you recommend as the backend for a real-time app, like a multiplayer game or a collaborative web app, where every client needs to see changes instantly? Answer from your own knowledge.Claude Opus 5.5 · no tools · excerpt

Don't use a database as the sync layer for games. Firestore-style latency and per-write cost will hurt you. Keep game state in process memory and persist only results.

It recommended Yjs with Liveblocks or PartyKit for collaborative apps, and Colyseus for browser games. For apps that are mostly records, it suggested Supabase Realtime, Convex or Firebase: "Convex is particularly pleasant for this."

That's the opposite of SpacetimeDB's idea, which is to put game state and logic in the database. When we described that idea without the name, the model picked SpacetimeDB at once:

Awareness · its pitch, not namedI want one system that is my database and my server logic at the same time, with clients subscribing to queries and getting live updates, typed client SDKs for TypeScript and game engines, and no separate API server to deploy. What would you use? Answer from your own knowledgeClaude Opus 5.5 · no tools · excerpt

SpacetimeDB matches what you described.

…Alternative if you didn't need game engines: Convex. It's the same idea (DB plus TypeScript server functions plus reactive queries, no API server) with a very polished web developer experience.

It correctly described how SpacetimeDB works, its client libraries for TypeScript, Unity and Unreal, BitCraft and the Maincloud service.

Asked "What is SpacetimeDB (spacetimedb.com)?", it gave an accurate description of the product as it was around its 1.0 launch, including the Clockwork Labs and BitCraft background. It didn't mention 2.0, the web push, C++ support or any agent tooling, and it called the product "open source" under a licence that "converts to a permissive license". SpacetimeDB's FAQ calls it source-available, and the AGPL it converts to isn't a permissive licence.

Then we asked about the new brand and the homepage tagline, and asked how to connect an agent to it over MCP:

Awareness · new name, with its taglineWhat do you know about Spacetime, the "reactive backend platform that AI agents love"? What does it do, and who is it for? Say clearly if you don't know.Claude Opus 5.5 · no tools · excerpt

I don't recognize that exact product or tagline, so I can't tell you reliably what it does or who it's for.

…That tagline sounds similar to how Convex markets itself: a reactive backend with TypeScript functions, often promoted as a good fit for AI coding agents.

It offered SpacetimeDB as "the closest match I'm aware of", and said it didn't know whether "Spacetime" was a rebrand. Awareness · MCP setupHow would I set up SpacetimeDB so my coding agent can build with it and inspect a live database over MCP? Give me the steps.Claude Opus 5.5 · no tools · excerpt

I'm not certain whether Clockwork Labs currently ships an official MCP server, so check their docs (spacetimedb.com/docs) for "MCP" or "AI" before building your own.

…I can write the small MCP server from step 5 in this sandbox if you'd like. It's about 60 lines of TypeScript using @modelcontextprotocol/sdk.

It suggested driving the CLI from a CLAUDE.md file, or writing a custom MCP server around the HTTP SQL endpoint. SpacetimeDB's own answer, the plugin and spacetime mcp, didn't come up.

The model didn't invent anything, and it said when it didn't know. Its picture of SpacetimeDB is the game backend from before 2.0.

OKAwareness3 / 4Knows the games database well; doesn't know the rebrand, the web push or the agent tooling

What SpacetimeDB can do about it

Training data catches up slowly, so the fix is content that teaches the next training cut about 2.0. Posts and guides that pair the name with the new pitch ("SpacetimeDB, a reactive backend for web apps and AI agents") will help, as will a short page explaining that "Spacetime" and "SpacetimeDB" are the same product. Using one name everywhere would remove the doubt the model had about the rebrand.

Discoverability: can an agent find SpacetimeDB?

Discoverability tests whether an agent with web search finds the product when it's looking for something like it. We ran three six-question "funnels", one session each. Each funnel starts with a broad problem and narrows towards SpacetimeDB one question at a time, then names it at question 6. We recorded the first question where SpacetimeDB appeared.

OK3 / 4Top pick for games; second to Convex for web appsFAILOnly appears when namedPOOROnly appears with its own positioning languageOKAppears for a specific description of the problemGOODTop choice for broad problem descriptions

Each funnel asks from a different angle. Funnel A is a web developer, the audience SpacetimeDB 2.0 is aimed at. Funnel B is a game developer, the audience it started with. Funnel C builds apps by prompting a coding agent, which is who its homepage now speaks to.

For game developers, SpacetimeDB was the answer from the second question to the last:

Discoverability · funnel B, question 2I don't want to write and host a separate game server plus a database. I want authoritative server logic, persistent state, and clients that stay in sync automatically. What fits?Claude Opus 5.5 · web search · excerpt

SpacetimeDB fits that description very closely. It combines the database and the game server in one system:

At question 3 it called SpacetimeDB "the only mature tool that meets all three of your requirements". Its one caveat throughout was fast-paced competitive games that need client-side prediction.

For the web app in funnel A, SpacetimeDB appeared at question 3, when we asked for server logic inside the database, ACID transactions and query subscriptions. The agent's opening line was "Only two tools meet all four of your requirements: Convex and SpacetimeDB. For your app I'd choose Convex." Convex stayed the pick until the end. Then question 5 used SpacetimeDB's homepage tagline, and the search found Convex instead:

Discoverability · funnel A, question 5Are there any products that describe themselves as a "reactive backend platform"?Claude Opus 5.5 · web search · excerpt

Yes. Convex is the only established product that uses the phrase "reactive backend platform", and it fits your requirements.

The top search result for the phrase was a third-party profile titled "Convex: Reactive Database and Backend Platform". At question 6, after reading spacetimedb.com, the agent said: "Yes, and I should have included it in my last answer."

Funnel C is the one SpacetimeDB's new homepage is written for. Here are its six questions, asked in one session, with the start of each answer.

1. "I build apps mostly by prompting Claude Code. What backend should I have it use so it doesn't keep breaking things?"No SpacetimeDB"Which backend you pick matters less than whether Claude can check its own work." Its first suggestion was Convex: "It was built with AI-written code in mind."
2. "My agent keeps getting lost wiring up a database, an API server, auth and WebSockets. I'd like one platform that covers all of that, so there's less glue for the agent to get wrong. What are the options?"No SpacetimeDB"For your case I'd use Convex." It also listed Supabase, Firebase, InstantDB, PocketBase and Appwrite.
3. "Ideally the platform has an official MCP server or agent skills, an llms.txt, and published evidence of how well LLMs write code for it. Which backends offer that? Search the web if you need to."Footnote"SpacetimeDB came up during the search. It publishes an LLM benchmark, but the only MCP server I found is a community one, and I didn't confirm an llms.txt." It recommended Convex, with Supabase "a close second".
4. "Look for newer or less well-known tools that do exactly this, including anything launched in the last few months. Search online."No SpacetimeDB"I found five newer platforms built for this." They were InsForge, Butterbase, Base44, Tencent's CloudBase and Cohesivity, and it still recommended Convex.
5. "Have you come across a backend that pitches itself as letting AI "ship the whole backend, glue-free"?"Not found"No, I haven't. I searched for the exact phrases "glue-free" and "ship the whole backend" and found no backend platform using either one."
6. "What about SpacetimeDB?"Found"Yes, SpacetimeDB is the one." After reading the homepage and agent-setup.md: "In fact it has an official MCP server and an llms.txt, so it meets all of your criteria." It still called Convex "the safer default".

Question 3 described SpacetimeDB's agent tooling almost item by item. The agent ran a search for "SpacetimeDB MCP server llms.txt agent skills AI", found a community MCP server and the benchmark page, and concluded the official tooling wasn't there. It only found the plugin, the MCP server and llms.txt after we named SpacetimeDB and it read the homepage. It also noticed a weakness of the benchmark: "The benchmark scores are at the ceiling. Opus 5.5 and two GPT-6 models all score 99.5%."

OKDiscoverability3 / 4Top pick for multiplayer games; second to Convex for web apps

What SpacetimeDB can do about it

Make the MCP server and agent skills findable by search

The agent searched for SpacetimeDB's MCP server, llms.txt and agent skills by name and found a community MCP server instead. A docs page titled in plain words, such as "Using SpacetimeDB with AI coding agents: MCP server, agent skills and llms.txt", would give that search an official result.

Put the taglines where search engines index them

Searches for "reactive backend platform" returned pages about Convex, and searches for "glue-free" and "ship the whole backend" returned nothing. The agent found both phrases on spacetimedb.com only after we named the product. Using them in page titles, the docs and launch posts would let a tagline search land on SpacetimeDB.

Win the web app recommendation

For web apps and agent-built apps, Convex was the agent's answer from the first or second question, before it had searched anything. The web articles it read were vendor round-ups of backends for AI coding agents, and SpacetimeDB wasn't in the ones it opened. Guides and comparisons written for web developers would give agents something to cite there.

Reputation: do agents recommend SpacetimeDB?

Reputation tests what the agent thinks once SpacetimeDB is named, including whether it recommends it over competitors and whether it gets the facts right. We ran seven questions, each in a fresh session with web search. Convex and Supabase were the two products the agent named most often before it found SpacetimeDB, five answers each, so we used them for the head-to-heads. (We've audited Supabase's own agent experience before, in Supabase vs PlanetScale.)

POOR2 / 4Recommended for games; loses to Convex and Supabase for web appsFAILNot recommended, or core facts wrongPOORHeavy caveats, facts unreliableOKRecommended for specific scenarios, facts mostly rightGOODRecommended over alternatives, facts right

Is SpacetimeDB good for web apps?

Four of the five verdict questions were about web apps, the market SpacetimeDB 2.0 is aimed at, and all four went against it. The agent kept SpacetimeDB for very high-frequency shared state and recommended Convex or Supabase for everything else. The one game question got a clear yes.

SpacetimeDB vs Convex and Supabase

Both head-to-heads used the same scenario, a real-time collaborative web app built in TypeScript and React. Here's the answer against Convex:

Reputation · SpacetimeDB vs ConvexWould you recommend SpacetimeDB over Convex for a real-time collaborative web app built in TypeScript and React?Claude Opus 5.5 · web search · excerpt

For most real-time collaborative web apps in TypeScript and React, I'd pick Convex. I'd choose SpacetimeDB only if your app has very high-frequency shared state, like a multiplayer game.

…Even SpacetimeDB's team says so. Their blog has a post titled "why you should choose Convex", which admits that speed isn't what matters most to most users and that Convex gets a lot right.

Its rule of thumb: "Something like Notion, Linear or Trello: Convex." SpacetimeDB was "worth a look" for something like Figma's live canvas or a multiplayer game. Reputation · a B2B SaaS dashboardWe're a 3-person team building a B2B SaaS dashboard in Next.js. We need Postgres-style reporting queries, Stripe webhooks, file uploads and email login. Is SpacetimeDB a good fit?Claude Opus 5.5 · web search · excerpt

Short answer: no. SpacetimeDB is good at one thing your app doesn't need much, and weak at the thing it needs most.

It read SpacetimeDB's SQL reference and called the limited SQL "the dealbreaker". It also said Stripe webhooks were unproven, and recommended Postgres on Neon or Supabase instead: "Supabase alone covers the DB, auth and storage". Reputation · an indie Unity studioWe're an indie studio making a multiplayer game in Unity with a few thousand concurrent players at launch. We need authoritative server logic, persistence and hosting we don't have to run ourselves. Should we shortlist SpacetimeDB?Claude Opus 5.5 · web search · excerpt

Yes, shortlist it, but first check one thing: does your authoritative logic need Unity on the server? That answer decides whether SpacetimeDB fits.

It recommended it for games driven by state, such as survival, strategy and social games, and pointed games that need server-side physics towards Photon.

The facts were mostly right, including when we used only the name. The pricing and licensing answer read SpacetimeDB's pricing page and its licence file and got every plan, price and licence term right. The answer on risks identified the product from its bare name, and its storage limits match SpacetimeDB's limits page. Here's how its main claims checked out. The two errors came from pages the agent read instead of SpacetimeDB's own.

Five of the seven answers cited the same unofficial mirror of SpacetimeDB's repository, and two took the out-of-date version number from it. The answer that compared SpacetimeDB with Supabase never opened spacetimedb.com. In our own check afterwards, we found a Spacetime vs Supabase comparison page on SpacetimeDB's site that it didn't reach.

POORReputation2 / 4Recommended for games; loses to Convex and Supabase for web apps

What SpacetimeDB can do about it

Answer the web app objections where agents look

Every web app verdict came back to the same few concerns, which were reporting queries, webhooks, file storage, auth and a smaller ecosystem. Some were wrong, such as file storage, and some were right, such as the SQL limits. A docs page that addresses each one for web developers, including how to build reporting with only COUNT in SQL, would give agents something to quote back.

Write comparisons that agents can use as a recommendation

The agent quoted SpacetimeDB's own Convex post as evidence against it ("Even SpacetimeDB's team says so."). A comparison page that names the web apps SpacetimeDB is the better choice for would give agents a reason to recommend it for those apps.

Keep the official sources ahead of copies

The mirror's out-of-date version number reached two answers. Release notes and the current version on pages that rank, such as the docs home and the GitHub README, would help the official sources win those searches.

Integration: can an agent build with SpacetimeDB?

Integration tests whether an agent can install the product and build a "hello world" with as little human help as possible. We gave it four CSV files for a fictional business (50 customers, 30 products, 300 orders with 742 line items) and asked it to:

1. Set up SpacetimeDB and get a local instance running.
2. Load the CSV data into a database and make sure it can query it.
3. Answer a question using the database: which region had the most revenue in the last three months, and who were its top three customers?
4. Build a web page that shows revenue per region live, then refund one order and prove the page updated without a reload.
5. Stop client apps reading customer email addresses, and prove it.

We ran it in Claude Code's default permission mode, approving each command, so we could stop the agent from reading anything outside its folder. The integration ran in a brand-new Claude Code profile, so SpacetimeDB's plugin and skills would have loaded normally if the agent had installed them.

GOOD4 / 4Live app with private data, correct answers and no human stepsFAILNo working hello worldPOORWorks, but needed several human stepsOKWorks with short human steps and minor workaroundsGOODEnd to end, headline feature included, at most one human step

Installing: four and a half minutes, no help

The agent knew which product we meant and how to install it. It downloaded the official install script, read the top of it, and ran it. It then started a local server, made the local server the CLI's default, and tested the setup with a throwaway test project, which it deleted afterwards.

Integration · step 1, installSet up SpacetimeDB on this machine and get a local instance running that I can build against.Claude Opus 5.5 · all tools · excerpt

Server is up (ping → 200). The CLI defaults to maincloud, so I'll switch the default to local so publishes don't go to the cloud by accident, then smoke-test with a real module.

SpacetimeDB 2.10.2 was installed and running in 4 minutes 28 seconds. There was no signup, licence screen or API key, and publishing to the local server logged the CLI in automatically.

Loading the data, guided by the CLI's rules file

The agent wrote server code that defines four tables, storing money as whole cents, and loader functions that refuse to load the same table twice. It then wrote a Python script that sent the CSV rows to those functions. The row counts matched the files. It found that order is a reserved word in SpacetimeDB's SQL and renamed the tables to plurals.

Its main reference for SpacetimeDB's TypeScript API was the rules file that spacetime init writes into every new project, as AGENTS.md, CLAUDE.md and Cursor rules. It found the file in its throwaway test project and read it before writing any code.

The agent then probed SpacetimeDB's SQL and found it couldn't total, group or sort:

Integration · step 2, load and queryThere are some CSV files in the data folder here. Create a SpacetimeDB database for this data, load it in, and make sure you can query it yourself so you can answer questions about it.Claude Opus 5.5 · all tools · excerpt

Querying: SpacetimeDB's own SQL handles filtering, joins and counting, but it can't add up, group or sort. Most real questions need those, like "revenue by region" or "top products". So I wrote shop-db/query.py: each time it runs, it reads the current data from SpacetimeDB into a temporary SQLite copy and answers there with full SQL.

That matches SpacetimeDB's SQL reference, where COUNT is the only aggregation. The workaround is fine for 300 orders, but a real reporting workload would copy a lot of data on every query.

The hello world: a correct answer

Integration · step 3, hello worldUsing the database, which region had the most order revenue in the last 3 months, and who were the top 3 customers in that region?Claude Opus 5.5 · all tools · excerpt

North had the most order revenue in the last 3 months: $212,621 from 61 orders. That's about 58% more than South, in second place at $134,658.

Its top three were Mateo Haddad, Ava Muller and Zara Dube. It stated its date window and that it excluded refunded orders, and checked the ranking didn't change under other definitions of revenue.

The answer matched the figures we calculated from the CSVs: $212,620.84 for North, excluding refunds. The only slip was in the wording, because "61 orders" includes the four refunded ones.

The live page: updated 23 ms after the refund

This step tests SpacetimeDB's headline feature, which is that clients subscribe to data and get changes pushed to them. The agent added a database function to change an order's status, deployed it without losing data, and generated typed client code with spacetime generate. It then built a small web page with SpacetimeDB's TypeScript client library that subscribes to the orders and customers tables.

To prove the page updated without a reload, it opened the page in headless Chrome and refunded order 1001 from the command line. Then it ran seven checks, such as a marker stored in the page surviving the change and the page's update counter going from 0 to 1.

Integration · step 4, live page and refundBuild a minimal web page that shows total order revenue per region and updates live. Then, through the database, mark order 1001 as refunded so it no longer counts towards revenue, and prove the page updated without a reload.Claude Opus 5.5 · all tools · excerpt

West dropped by exactly $6,425.19, the value of order 1001. The page updated about 23 ms after the database change.

It reported one test run it couldn't explain, an extra browser navigation event even though the page hadn't reloaded, and said it had narrowed that check. It also warned that its new function "has no permission check".

Both figures match our own: West went from $225,607.88 to $219,182.69.

Hiding customer emails: a private table and a public view

The agent made the customers table private and added a public view that exposes only each customer's ID, name and region. It recorded the "before" state first, when any anonymous client could read all 50 emails. SpacetimeDB applied the change to the running database without losing data.

Integration · step 5, hide emailsClient apps connecting to this database must not be able to read customer email addresses, but should still see customer names and regions. Make that so, and prove it.Claude Opus 5.5 · all tools · excerpt

Plan: make the customers table private, so only the database owner (my CLI identity) can read it. Then add a public view, customer_public, that exposes only customer_id, name and region. Client apps read the view; the email column never leaves the server.

It proved it four ways: from the command line as an anonymous client, over the HTTP API as a new identity, from a second client in the browser, and by decompressing every message the browser received and searching for all 50 email addresses. None were there.

We checked it ourselves as an anonymous client. Reading the customers table returned "no such table: customers. If the table exists, it may be marked private.", and the view returned names and regions without emails.

What it used, and what it didn't

Finally, we asked which SpacetimeDB docs and agent tools it had used:

Integration · step 6, toolingWhich SpacetimeDB docs, MCP server, agent skills, plugins, SDKs, or llms.txt did you use to do all this? Did you know about them before you started?Claude Opus 5.5 · all tools · excerpt

I didn't use any SpacetimeDB docs website, llms.txt file, MCP server, Claude Code skill or plugin.

…Didn't know: that spacetime init creates AGENTS.md, CLAUDE.md and Cursor rule files. I came across them in the throwaway test project during setup and relied on them afterwards.

The rest came from its training data, the spacetimedb npm package's own type definitions and source code, and trial and error against the running server. It never searched the web during the session.

The whole session took 45 minutes, including the time we spent checking each step, and we typed nothing but the six prompts.

GOODIntegration4 / 4Live app with private data, correct answers and no human steps

What SpacetimeDB can do about it

Point the rules file at the rest of the agent tooling

The rules file that spacetime init writes was the only piece of SpacetimeDB's agent tooling the agent used. A short section in that file on the plugin, spacetime mcp and agent-setup.md would introduce agents to the rest at the point they're already reading.

Document the reporting pattern

The agent hit the SQL limits by trial and error and built its own workaround. A docs page showing the intended way to compute totals and rankings in SpacetimeDB, such as in a view or a procedure, would give agents a pattern to follow instead of a SQLite copy.

What this means for SpacetimeDB

SpacetimeDB is a good product for an agent to build on. Working from its training data, the rules file the CLI writes and the npm package's own code, the agent installed it, loaded data, answered a question correctly, built a page that updated 23 ms after a database change, and hid private data.

The problems come before an agent starts building:

• From memory, agents know SpacetimeDB as a game backend, and its new tagline sounds like Convex to them.
• For web apps, agents recommend Convex or Supabase. SpacetimeDB lost both head-to-heads and the SaaS question.
• Agents searching for its MCP server, skills and llms.txt found a community MCP server instead.

Guides for web developers, comparison pages that recommend SpacetimeDB for specific kinds of app, and an agent tooling page that ranks for those searches would each address one of these.

Want help understanding and fixing the AX of your own technical product or platform? At Ritza our Engineering Writers work at agent speed with human-expert verification (no slop) to win at GTM.