fix: minor changes to the tagging prompts (#1474)

* feat: add optional `thinking` key to tagging response schema

* prompt: fix indent

* prompt: remove extra 'language' word

* prompt: use xml as separator

* revert: dont use a thinking tags

Signed-off-by: thiswillbeyourgithub
<26625900+thiswillbeyourgithub@users.noreply.github.com>

* prompt: don't ask to include website tags

* prompt: aim for 5 tags

* prompt: dont tell bot its a bot

* prompt: propose a tag_error

* Revert "prompt: propose a tag_error"

This reverts commit 78c5099a187960cc3697b77f2b2bd687edb015f3.

* minor prompt tweaks

* minor prompt tweaks take 2

---------

Signed-off-by: thiswillbeyourgithub
Co-authored-by: Mohamed Bassem <me@mbassem.com>

1 files changed, 11 insertions, 10 deletions

diff --git a/packages/shared/prompts.ts b/packages/shared/prompts.ts
index 7ee6f12a..6928ec3e 100644
--- a/packages/shared/prompts.ts
+++ b/packages/shared/prompts.ts
@@ -21,10 +21,10 @@ function truncateContent(content: string, length: number) {
 
 export function buildImagePrompt(lang: string, customPrompts: string[]) {
   return `
-You are a bot in a read-it-later app and your responsibility is to help with automatic tagging.
+You are an expert who's responsibility is to help with automatic text tagging for a read-it-later app.
 Please analyze the attached image and suggest relevant tags that describe its key themes, topics, and main ideas. The rules are:
 - Aim for a variety of tags, including broad categories, specific keywords, and potential sub-genres.
-- The tags language must be in ${lang}.
+- The tags must be in ${lang}.
 - If the tag is not generic enough, don't include it.
 - Aim for 10-15 tags.
 - If there are no good tags, don't emit any.
@@ -40,18 +40,19 @@ export function buildTextPrompt(
 ) {
   content = preprocessContent(content);
   const constructPrompt = (c: string) => `
-You are a bot in a read-it-later app and your responsibility is to help with automatic tagging.
-Please analyze the text between the sentences "CONTENT START HERE" and "CONTENT END HERE" and suggest relevant tags that describe its key themes, topics, and main ideas. The rules are:
+You are an expert who's your responsibility is to help with automatic tagging for a read-it-later app.
+Please analyze the TEXT_CONTENT below and suggest relevant tags that describe its key themes, topics, and main ideas. The rules are:
 - Aim for a variety of tags, including broad categories, specific keywords, and potential sub-genres.
-- The tags language must be in ${lang}.
-- If it's a famous website you may also include a tag for the website. If the tag is not generic enough, don't include it.
+- The tags must be in ${lang}.
+- If the tag is not generic enough, don't include it.
 - The content can include text for cookie consent and privacy policy, ignore those while tagging.
 - Aim for 3-5 tags.
 - If there are no good tags, leave the array empty.
 ${customPrompts && customPrompts.map((p) => `- ${p}`).join("\n")}
-CONTENT START HERE
+
+<TEXT_CONTENT>
 ${c}
-CONTENT END HERE
+</TEXT_CONTENT>
 You must respond in JSON with the key "tags" and the value is an array of string tags.`;
 
   const promptSize = calculateNumTokens(constructPrompt(""));
@@ -67,9 +68,9 @@ export function buildSummaryPrompt(
 ) {
   content = preprocessContent(content);
   const constructPrompt = (c: string) => `
-    Summarize the following content responding ONLY with the summary. You MUST follow the following rules:
+Summarize the following content responding ONLY with the summary. You MUST follow the following rules:
 - Summary must be in 3-4 sentences.
-- The summary language must be in ${lang}.
+- The summary must be in ${lang}.
 ${customPrompts && customPrompts.map((p) => `- ${p}`).join("\n")}
     ${c}`;