# Chrome Prompt API Hacker Approved page: https://hackerapproved.com/p/chrome-prompt-api/ 8 people recommended Chrome Prompt API in Hacker News comments, 6 had mixed feelings and 23 criticized it, November 2025 to October 2026. It was recommended in 4 of 12 months, in 4 threads. It does not meet the Hacker Approved badge rule. 4 of the 8 people who recommend it commented in one thread, “The Prompt API”. **Verdict:** More people criticize it than recommend it. Free, on-device AI for web pages with no server costs or quotas, suited to privacy-focused and structured-output tasks. Main caveat: fingerprinting and multi-gigabyte model downloads. Summary written by Claude from the comments below. ## What people warn about - Fingerprinting and privacy risk (2 comments in the sample): "This will be a new source of fingerprinting information and this is difficult to fake to fool fingerprinting scripts" ([source](https://news.ycombinator.com/item?id=47963299)) "This outright non-feature is a significant source of issues, least of which is fingerprinting." ([source](https://news.ycombinator.com/item?id=47966237)) - Web scope and precedent concerns (2 comments in the sample): "We don't need the browser to have a Prompt API to enable web applications to have goofy chatbots lurking in the corner." ([source](https://news.ycombinator.com/item?id=47961323)) "It is non-obvious that adding an LLM to a web browser makes anything better for the web browser user." ([source](https://news.ycombinator.com/item?id=47968275)) - Inconsistent behavior across browsers and models (2 comments in the sample): "when each browser has a different model attached to this API testing becomes even more a nightmare" ([source](https://news.ycombinator.com/item?id=47920119)) "a model update will work entirely differently and break the existing implementations" ([source](https://news.ycombinator.com/item?id=47963805)) - Model too limited for longer tasks (2 comments in the sample): "that is ~9% of the total available disk space for baseline phones and laptops for a model that is not that useful" ([source](https://news.ycombinator.com/item?id=47918037)) "The model this uses is useless for anything beyond 2 round chat at the most." ([source](https://news.ycombinator.com/item?id=47917499)) ## Why people recommend it - Structured JSON output and tool calling (3 comments in the sample): "leading to model quality improvements, language support expansion, API additions like structured responses and tool use" ([source](https://news.ycombinator.com/item?id=47962180)) "if you have a use case that needs constrained JSON output for example, it's pretty neat" ([source](https://news.ycombinator.com/item?id=48088283)) "particularly because it has tool-calling builtin & structured output" ([source](https://news.ycombinator.com/item?id=47918218)) - Free, no server costs or quotas (2 comments in the sample): "the upsides of on-device (e.g. zero-cost, permission-less start with no quotas/infra, network-resilience, privacy) make it a compelling trade-off" ([source](https://news.ycombinator.com/item?id=47930797)) "The main win is that it's free and privacy preserving" ([source](https://news.ycombinator.com/item?id=47917453)) - Small reasoning on user data (2 comments in the sample): "I've integrated these into my workflows where a tiny but non zero amount of reasoning is needed" ([source](https://news.ycombinator.com/item?id=48366491)) "it shines in constrained environments where the model is transforming user-owned data" ([source](https://news.ycombinator.com/item?id=48088205)) ## The badge A product earns the Hacker Approved badge when at least 10 people recommend it, recommenders outnumber critics at least 2 to 1 (mixed feelings count toward neither side), it was recommended in at least 4 of the 12 months and in at least 5 different threads, and no single thread supplies more than half of the people who recommend it. Health and money products are never badged. ## Cite as Hacker Approved, "Chrome Prompt API", https://hackerapproved.com/p/chrome-prompt-api/ (8 recommend, 6 mixed, 23 criticize; Hacker News comments, November 2025 to October 2026). Counts updated 2 Nov 2026, with data through October 31, 2026.