feat(server): update tool descriptions and AI prompt (#13032)

update tools description & chat prompt

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **New Features**
* Updated the AFFiNE AI copilot system prompt to reflect support for
multiple AI providers and a more concise, structured format with clearer
guidelines and modular tags.

* **Enhancements**
* Improved descriptions for document search and reading tools, providing
clearer guidance on when and how to use keyword search, semantic search,
and document reading features.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

packages/backend/server/src/plugins/copilot/tools/doc-keyword-search.ts

@@ -39,7 +39,7 @@ export const createDocKeywordSearchTool = (
 ) => {
   return tool({
     description:
-      'Full-text search for relevant documents in the current workspace',
+      'Search all workspace documents for the exact keyword or phrase supplied and return passages ranked by textual match. Use this tool by default whenever a straightforward term-based lookup is sufficient.',
     parameters: z.object({
       query: z.string().describe('The query to search for'),
     }),

packages/backend/server/src/plugins/copilot/tools/doc-read.ts

@@ -72,7 +72,8 @@ export const createDocReadTool = (
   getDoc: (targetId?: string) => Promise<object | undefined>
 ) => {
   return tool({
-    description: 'Read the content of a doc in the current workspace',
+    description:
+      'Return the complete text and basic metadata of a single document identified by docId; use this when the user needs the full content of a specific file rather than a search result.',
     parameters: z.object({
       doc_id: z.string().describe('The target doc to read'),
     }),

packages/backend/server/src/plugins/copilot/tools/doc-semantic-search.ts

@@ -56,7 +56,7 @@ export const createDocSemanticSearchTool = (
 ) => {
   return tool({
     description:
-      'Semantic search for relevant documents in the current workspace',
+      'Retrieve conceptually related passages by performing vector-based semantic similarity search across embedded documents; call this tool only when exact keyword search fails or the user explicitly needs meaning-level matches (e.g., paraphrases, synonyms, broader concepts).',
     parameters: z.object({
       query: z
         .string()