WhoDB and SQLly: a browser client you host and a desktop SQL workbench
WhoDB is an Apache-2.0 Go server you open in a browser, with MongoDB and Elasticsearch; SQLly is a native Rust workbench with SQL Server, Oracle and deep SQL tooling. Where each one wins.
WhoDB is an
open-source database client from Clidey, a Y Combinator company, with about 5,000 stars on GitHub
and a new release most weeks. Its README calls it “a lightweight, self-hosted workspace for
your databases”. You start it with one docker run, open a browser tab, and
you’re browsing tables. SQLly takes the opposite route: a native desktop app with no
browser engine at all. This post sets the two side by side.
We build SQLly, so take the bias into account. We read WhoDB’s README, its documentation, its pricing and integrations pages, and the sizes of its 0.138.0 release assets (28 October 2026). We did not install it or benchmark it. For several kinds of user it is the better choice, and we say which ones below.
A server you open in a browser, or an app on your desk
WhoDB is a Go backend serving a React and TypeScript front end. That one design gives you several ways to run it:
- In Docker or as a binary.
docker run -p 8080:8080 clidey/whodb, then openlocalhost:8080. The image is about 73 MB compressed on amd64, and there are server binaries for amd64, arm64, armv6, armv7 and riscv64. - As a desktop app. A Wails app that shows the same front end in the system web view, on the Mac App Store, the Microsoft Store and Snap. The Mac build is universal, so it runs on Intel.
- In a terminal. A separate CLI with a full-screen terminal UI, an MCP server, cloud discovery for AWS, Azure and GCP, and JSON or CSV output for scripts.
The browser version is the real win. Put WhoDB on one server and a whole team can explore data from any machine, without installing anything. SQLly can’t do that. It is a single Rust codebase drawn on the GPU through Metal, Direct3D or Vulkan, and you install it on each machine. Our browser preview is a demo, not something you host for colleagues.
The trade-off is on the desk. WhoDB’s downloads are 139.4 MB for the Mac dmg (83.1 MB on
the Mac App Store), 132.9 MB for the Windows exe and 136.8 MB for the Linux server binary. SQLly’s
are 65.6 MB, 60.9 MB and 76.2 MB, though ours are compressed archives and theirs mostly aren’t,
so don’t read too much into it. Hosting also brings chores a desktop app doesn’t have:
WhoDB’s README tells you to generate and keep an encryption key so sessions survive a
container restart, and to set WHODB_SECURE=true behind an HTTPS proxy.
Databases
| Engine | SQLly | WhoDB Community |
|---|---|---|
| PostgreSQL, MySQL, MariaDB, SQLite | Yes | Yes |
| DuckDB, ClickHouse | Yes | Yes (DuckDB not on every architecture) |
| Redis, Valkey, DragonflyDB | Yes, plus KeyDB, ElastiCache and MemoryDB | Yes |
| CockroachDB and other Postgres/MySQL compatibles | Documented (Neon, Supabase, Aurora…) | CockroachDB, YugabyteDB, QuestDB, TiDB |
| SQL Server, Azure SQL | Yes | No (commercial Platform only) |
| Oracle | Yes | No (commercial Platform only) |
| libSQL/Turso | Yes | No |
| MongoDB, FerretDB | No | Yes |
| Elasticsearch, OpenSearch | No | Yes |
| Memcached | No | Yes |
The split is clean. WhoDB’s free edition reaches into document stores, search engines and caches, which SQLly doesn’t touch. SQLly covers the commercial SQL engines, SQL Server, Azure SQL and Oracle, which WhoDB keeps for its paid Platform. If your day job is SQL Server, that one row probably decides it. The engine list shows which SQLly tools work on which engine.
Writing SQL
WhoDB’s Scratchpad is a set of query cells, each with its results underneath, plus syntax highlighting, table and column autocomplete, run-selection and history. It asks for confirmation before destructive SQL. That covers the everyday work.
SQLly puts most of its effort here. IntelliSense
tracks aliases and scope, writes a JOIN ON clause from real foreign keys, pulls a
column’s actual values into a WHERE clause, turns “last Tuesday”
into a date range, and flags a GROUP BY mistake while you type. The editor also has
parameters, templates, multi-cursor editing, split views, folding and AI ghost text, and a
half-megabyte script edits at about 5 ms a keystroke.
Results
WhoDB’s grid feels like a spreadsheet: sort, filter with a visual WHERE
builder, page through rows, edit inline, add rows in a dialog. It exports CSV, Excel and NDJSON,
and imports CSV, Excel or a SQL script up to 50 MB and 100,000 rows. Its mock-data generator
fills a table with up to 200 rows at a time and respects foreign keys. For pivots and charts its
docs point you to Excel.
SQLly’s grid streams rows to disk instead of paging, so fifty million rows use about as much memory as fifty. On top of it are pivots, charts, conditional formatting, computed columns, result diffs and Excel export with live formulas and a pivot sheet, and a one-line comment saves the layout with the query.
AI and MCP
Both apps have an AI chat that can see your schema. WhoDB supports Ollama and LM Studio locally, and OpenAI, Anthropic, Gemini or any OpenAI-compatible service with your key. Its chat can also insert, update, delete and change the schema, after showing you the SQL and waiting for you to confirm. SQLly supports local models through Ollama and MLX, your own endpoint, or opt-in cloud providers, with no provider switched on by default; see on-device AI.
Both ship an MCP server, and here the two are closer than we expected. WhoDB’s
documentation describes a shared SQL classifier and native read-only transactions for reads on
the Postgres and MySQL families, a read-only file connection on SQLite, and
readonly=1 on ClickHouse. That is the same idea SQLly uses. The difference is
what happens with writes:
- WhoDB lets agents write once you confirm.
whodb mcp serveallows reads and confirmed writes by default.--read-onlyblocks writes,--safe-modeadds strict checks, and--allow-writeskips the confirmation for trusted automation.--allowed-connectionslimits which connections an agent can see. - SQLly never lets an agent write.
sqlly-cli mcplets an agent list connections, read schemas and run oneSELECTat a time in a read-only transaction that is always rolled back. If the agent wants to change data, it sends a proposal; SQLly shows you the exact SQL and you run it or decline it. Production connections are refused outright. Details are in how reads are kept read-only.
Safety and admin tools
WhoDB’s guardrails are mostly about confirmations: before destructive SQL in the scratchpad, before AI chat writes, before MCP writes. It also has an optional append-only audit log for self-hosted deployments.
SQLly’s guardrails go further for the person at the
keyboard: SELECT-only connections that block file-reading and shell-escape tricks, a hard stop
on UPDATE or DELETE without WHERE, automatic rollback,
tenant checks, an optional AI review before a data change, and production painted red. You set
them on a client, project or environment and every connection under it inherits them.
For admin work, WhoDB has a schema graph, a schema diff and native EXPLAIN in the
CLI, and you run EXPLAIN in a scratchpad cell on the web. We couldn’t find an
activity monitor; its DBA guide has you query pg_stat_activity yourself. SQLly has
the full set under one Tools menu: Activity Monitor with
blocking chains, Server Dashboard, Query Insights, Extended Events and SQL Agent for SQL Server,
query plans on nine engines (still in progress), an
index analyzer, risk-graded schema compare
with rollback scripts, data compare, and cross-engine
Data Transfer.
What they cost
WhoDB Community is free, Apache 2.0, and fine to self-host at any company size. Its enterprise code sits in a separate private repository. The whodb.com site mostly sells WhoDB Platform, a hosted data operations product with projects, pipelines, lineage and SSO, priced per member: Free for 2 members, Start at $20, Pro at $200 per member per month, and Enterprise on request. That Platform is where SQL Server, Oracle, Snowflake and BigQuery live.
SQLly has a free tier for hobbyists (two saved servers), a one-time $100 SQLly 1.x license and a $200 Lifetime license; see pricing. If “open source” or “free for the whole company” is a requirement, WhoDB Community meets it and SQLly does not. One more thing worth knowing: WhoDB’s web UI turns on anonymous usage data, including session replay, by default and shows a banner where you can switch it off.
Which one to pick
Pick WhoDB if:
- You want one install that a whole team opens in a browser.
- You need MongoDB, Elasticsearch or Memcached.
- You need open source, or free at any company size.
- You’re on an Intel Mac, a Raspberry Pi or RISC-V, or you want a terminal UI.
- You want the interface in one of its 60-plus languages.
- You want an AI agent to change data after you confirm it.
Pick SQLly if:
- You work in SQL Server, Azure SQL, Oracle or libSQL.
- You write a lot of SQL and want completions that know your data.
- You pull large result sets, or pivot and chart them in place.
- You want guardrails on your own queries that you set once and inherit everywhere.
- You read plans, compare schemas, or watch server activity.
The fair test is your own database. WhoDB is one docker run away and SQLly installs
in a minute; neither needs an account to start. The side-by-side table is at
SQLly vs WhoDB, and SQLly’s download is free.