Comparison · Tool comparison
How does Ritoko compare with taprun for saved browser tasks?
Short answer
Both save browser tasks for local replay without a live model choosing every action. Ritoko focuses on CSV/XLSX rows, explicit business keys and a local recovery journal. taprun documents drift verification, script converters and uncertain-intent resolution. Check its engine licensing and exact recovery semantics before adopting it.
Updated · Sources checked
↔ Scroll to read all columns.
| Decision | Ritoko | taprun |
|---|---|---|
| Saved representation | JSON workflow | JSON plan |
| Browser path | Personal Chrome or explicitly chosen clean profile | README describes Chrome extension and Playwright paths; check version |
| Drift and maintenance | Fallbacks and deliberate supported selector repair | Documented verify verdicts and re-capture path |
| Uncertain write | Review item resolved after a destination check | Documented intent_uncertain and mark resolution |
| Spreadsheet recovery | CSV/XLSX input and durable row outcomes | Equivalent CSV/XLSX crash-resume contract not established by sources read |
| Engine licence | MIT repository and runtime | Public statements conflict; obtain clarification for the exact binary |
taprun documentation is not an independent engine audit. Different vendor sources describe different runtime and licence scopes.
What do the two approaches share?
taprun’s README describes capture into a saved plan, deterministic replay, local browser paths, drift checks and converters from Playwright, Puppeteer and Stagehand scripts. It is a relevant alternative for recurring browser work, not a tool that necessarily reasons through every repeated action. Source: taprun README.
Ritoko likewise separates discovering a task from executing supported saved steps. In both cases, inspect the actual task contract and the work still needed for authoring, maintenance and verification.
When might taprun be the more useful choice?
Consider it when its plan ecosystem, existing-script converters or explicit drift-check command fit your task. Its vendor index documents verify verdicts and mark to resolve an intent_uncertain run as committed or aborted. It would be unfair to claim that taprun has no uncertain-write handling. Source: vendor tool index.
Before a production write batch, verify what a key represents, how long any deduplication state lasts, whether uncertainty blocks later submissions and how state survives a crash. The sources read do not establish a matching spreadsheet row-resume contract.
When is Ritoko’s spreadsheet contract a useful fit?
Choose Ritoko when the input is CSV/XLSX and the job is a known, verifiable operation per row. A workflow declares the key, destination scope, one irreversible commit and result checks. Its SQLite journal freezes the workflow and rows, normally skips confirmed keys and holds uncertain writes across later runs.
Direct replay contains no LLM call, but recording, repair, host orchestration and model-backed external tools can consume model usage. Ritoko’s evidence is limited too: controlled demos and tests do not establish reliability on every site, and its journal cannot track independent submissions.
What remains uncertain about taprun’s licence?
The live homepage calls taprun MIT-licensed, while the public packages README describes the consuming Tap CLI as proprietary and closed-source. The main README limits its MIT statement to the extension and docs. These are conflicting primary-source statements about different parts of the distribution.
We therefore do not label the engine MIT, AGPL or proprietary as a settled conclusion. Check the exact binary’s terms with the maintainer before relying on a licence assumption. Sources: live homepage and public packages and engine boundary. No vendor benchmark or claimed cost advantage is treated as a measured comparison with Ritoko.