@7mioda · Paris, France
Edaly Ahmed
Building quietly: chaos, code, and a sparkle of AI ✨
- Merged PRs
- 944
- Reviews
- 5,067
- Repos
- 12
- Followers
- 33
- Merged
- 94%
tomorro/aura#372The input/output paths here look swapped. `editExecutor.execute(outputPath, inputPath, ...)` — you're passing `outputPath` as the source and `inputPath` as the destination. ShouldnAPR 2026tomorro/aura#372This `buildPdfFilename` method is never called — looks like dead code. Was this supposed to replace the inline filename logic in `getPreview`?APR 2026tomorro/aura#372The `activeDocuments` map has no TTL or max-size. If `cleanupThread` fails or the observable errors before `finalize()` runs, these entries leak. Is there a plan for this? Even a sAPR 2026tomorro/aura#372The `execute` method signature takes `docxPath` and `outputPath` as separate params but in the worker it's called with them swapped (see worker comment). We should double-check theAPR 2026tomorro/aura#372The `nonEditEvents` filter casts `event.data` to `AiMessage | undefined` and checks `msg?.id`. Is the `editMessageIds` set actually being populated before `onComplete` fires? The sAPR 2026tomorro/aura#372This shouldn't be committed. Add it to `.gitignore`.APR 2026tomorro/aura#372The `openSafeZipFromPath` function is async but `SafeZip` constructor is sync — the async is just for the `fs.readFile`. That's fine, but worth noting that `openSafeZip(buffer)` isAPR 2026tomorro/aura#372The worker is still using `fs.mkdtempSync`, `fs.writeFileSync`, `fs.readFileSync`, `fs.rmSync` — all sync. These block the worker thread. Should be `fs.promises.*` throughout.APR 2026tomorro/aura#372Are these docs meant to live in the repo permanently? The action-plan, concerns, poc-2-flow-analysis, poc-2-implementation-plan — these feel like working documents. Happy to keep tAPR 2026tomorro/aura#372We should remove these `console.error` calls. Either let the error propagate (the catch already returns `{ success: false, error: String(err) }`) or use the NestJS `Logger`. Same fAPR 2026tomorro/aura#445the bugbot is right — you need to `setStatusMessage(null)` here in the non-`status_update` branch so a stale server message doesn't permanently override the `actionKey` label. sameAPR 2026tomorro/aura#445if `label` is an empty string this renders nothing, but the logo still shows — is that intentional? feels like it might be worth falling back to a non-shimmer empty span so the layAPR 2026leafine-labs/earth#2002you can just spread `w` into the `HStack` via `...rest` instead of extracting it separately — unless there's a reason it needs to be excluded from the recipe variants?APR 2026leafine-labs/earth#2003why `typeof window === 'undefined'` instead of `typeof document === 'undefined'`? both work but we use the `document` check elsewhere for SSR detectionAPR 2026leafine-labs/earth#2003the `saveDraft` on every values change with no debounce could get noisy — maybe wrap it in a debounce?APR 2026leafine-labs/earth#2003the `!value.id.startsWith('temp-')` check is dead code now since you changed the id format to `import-address-${...}` on line 262 — you can remove this conditionAPR 2026leafine-labs/space#1344you might want to use `?.` here instead of the ternary, just a bit cleaner — `data.shippingAddress ? { ...data.shippingAddress, location: null } : null` works but `data.shippingAddAPR 2026tomorro/aura#422maybe an env variableAPR 2026tomorro/aura#429mayebe a wrapper for google tag manager in a separate react componentAPR 2026tomorro/aura#429I would put the try catch inside the functionAPR 2026