SMF·GRIST
Can a prompt replace Grist?
Databases — spreadsheet databases and operational data apps
Exhibit tracking slip
Verdict
Grist's real distinguishing feature — formulas written in actual Python, not a proprietary spreadsheet language — is genuinely a fun, buildable weekend project if you're comfortable running a sandboxed Python execution step per formula cell. Grist itself is also open source, so self-hosting the real project remains the more complete option. What a from-scratch build gives up either way: real-time collaboration and the safety hardening a maintained sandbox needs over years.
Exhibit A — The prompt
Received on31.07.2026Build a spreadsheet-database tool where formula columns are written in real Python, not a proprietary formula language — Grist's actual distinguishing idea. Use Next.js/TypeScript for the UI and a sandboxed Python execution step (run formula code in a restricted subprocess, or a WASM-based Python runtime like Pyodide for stronger isolation) for evaluating formula cells against the row's other column values. Support text, number, boolean, date, and formula column types in a table; a formula column's Python expression can reference other columns by name and recalculates when a dependency changes. Render the table as an editable grid. Add CSV import/export and basic dependency-cycle detection so a formula referencing itself fails clearly instead of hanging. Be explicit in the README about the sandbox's actual limits — resource limits on CPU, memory, and time, no network access from formula code, and that this hasn't had a security audit, since running arbitrary user-written Python safely is a genuinely hard, ongoing problem. Do not build real-time collaboration or a public sharing/API layer — those are out of scope. If you'd rather not build and maintain the sandbox yourself: Grist's own core is open source and self-hostable via Docker, which is the safer choice for anything beyond a personal, trusted-user project. Requires hosting; no external API key needed.Opening prefills the prompt — press enter to run it.
Exhibit B — What you lose
- B.1 real-time multiplayer editing
- B.2 a maintained, hardened formula sandbox
- B.3 the growing integration and API surface
- B.4 managed hosting and backups
Prior art
Exhibit C — Why people still pay: data model flexibility, collaboration, and integrations
A Python formula engine sounds simple; safely sandboxing arbitrary user-written Python so it can't escape and touch the host system, for years, without a security incident, is the unglamorous hard part.
Questions
Is it safe to run arbitrary Python formulas?
Only within real limits this build sets — resource caps, no network access — and even then, sandboxing arbitrary code safely is genuinely hard. For anything beyond your own personal, trusted use, self-hosting the real Grist is the safer choice.
Why not just self-host Grist itself?
That's the better answer for most people, since it's open source and already has a maintained, hardened sandbox. This build is for understanding the idea, or wanting something smaller and fully yours.
Can formulas reference other tables, not just other columns in the same row?
Not in this build — cross-table formula references are meaningfully more complex and are out of scope here.
What does it cost to run?
Hosting only — typically a few dollars a month, since there's no external API this build depends on.
Related tools
Receipt