gnosisllm-knowledge 0.3.0__py3-none-any.whl → 0.4.3__py3-none-any.whl

gnosisllm_knowledge/backends/opensearch/setup.py

@@ -249,10 +249,9 @@ class OpenSearchSetupAdapter:
             self._model_id = self._config.model_id
 
         # Step 4: Create ingest pipeline
-        # Only create ingest pipeline for global setup (not per-account)
-        # Account indices should use the global pipeline to ensure consistent model
-        is_global_setup = self._config.index_prefix == "gnosisllm"
-        if self._model_id and is_global_setup:
+        # Create pipeline for any setup that has a model deployed
+        # Each index_prefix namespace gets its own pipeline
+        if self._model_id:
             try:
                 await self._create_ingest_pipeline()
                 pipeline_name = self._config.ingest_pipeline_name or f"{self._config.index_prefix}-ingest-pipeline"
@@ -261,35 +260,33 @@ class OpenSearchSetupAdapter:
                 errors.append(f"Failed to create ingest pipeline: {e}")
                 logger.error(f"Failed to create ingest pipeline: {e}")
 
-        # Step 5: Create search pipeline (only for global setup)
-        if is_global_setup:
-            try:
-                await self._create_search_pipeline()
-                pipeline_name = self._config.search_pipeline_name or f"{self._config.index_prefix}-search-pipeline"
-                steps_completed.append(f"Created search pipeline: {pipeline_name}")
-            except Exception as e:
-                errors.append(f"Failed to create search pipeline: {e}")
-                logger.error(f"Failed to create search pipeline: {e}")
+        # Step 5: Create search pipeline for hybrid search
+        try:
+            await self._create_search_pipeline()
+            pipeline_name = self._config.search_pipeline_name or f"{self._config.index_prefix}-search-pipeline"
+            steps_completed.append(f"Created search pipeline: {pipeline_name}")
+        except Exception as e:
+            errors.append(f"Failed to create search pipeline: {e}")
+            logger.error(f"Failed to create search pipeline: {e}")
 
-        # Step 6: Create index template (only for global setup)
-        # Template covers all gnosisllm-* indices including per-account indices
-        if is_global_setup:
-            try:
-                template_name = f"{self._config.index_prefix}-template"
-                template_body = get_index_template(self._config)
+        # Step 6: Create index template for this namespace
+        # Template covers all {index_prefix}-* indices
+        try:
+            template_name = f"{self._config.index_prefix}-template"
+            template_body = get_index_template(self._config)
 
-                # Ensure template has global pipeline for auto-index creation
-                global_pipeline = self._config.ingest_pipeline_name or "gnosisllm-ingest-pipeline"
-                template_body["template"]["settings"]["index"]["default_pipeline"] = global_pipeline
+            # Set default pipeline for auto-index creation within this namespace
+            default_pipeline = self._config.ingest_pipeline_name or f"{self._config.index_prefix}-ingest-pipeline"
+            template_body["template"]["settings"]["index"]["default_pipeline"] = default_pipeline
 
-                await self._client.indices.put_index_template(
-                    name=template_name,
-                    body=template_body,
-                )
-                steps_completed.append(f"Created index template: {template_name}")
-            except Exception as e:
-                errors.append(f"Failed to create index template: {e}")
-                logger.error(f"Failed to create index template: {e}")
+            await self._client.indices.put_index_template(
+                name=template_name,
+                body=template_body,
+            )
+            steps_completed.append(f"Created index template: {template_name}")
+        except Exception as e:
+            errors.append(f"Failed to create index template: {e}")
+            logger.error(f"Failed to create index template: {e}")
 
         # Step 7: Create knowledge index
         try:
@@ -298,9 +295,8 @@ class OpenSearchSetupAdapter:
 
             if not exists:
                 settings = get_knowledge_index_settings(self._config)
-                # Add default pipeline - always use global pipeline for consistency
-                # This ensures all accounts use the same embedding model
-                pipeline_name = self._config.ingest_pipeline_name or "gnosisllm-ingest-pipeline"
+                # Add default pipeline for this namespace
+                pipeline_name = self._config.ingest_pipeline_name or f"{self._config.index_prefix}-ingest-pipeline"
                 settings["index"]["default_pipeline"] = pipeline_name
 
                 await self._client.indices.create(

gnosisllm_knowledge/cli/app.py

@@ -1,6 +1,11 @@
 """GnosisLLM Knowledge CLI Application.
 
 Main entry point assembling all CLI commands with enterprise-grade UX.
+
+Note:
+    This library is tenant-agnostic. Multi-tenancy is achieved through index
+    isolation - each tenant should use a separate index (e.g., "knowledge-{account_id}").
+    Use --index to target tenant-specific indices.
 """
 
 from __future__ import annotations
@@ -147,17 +152,13 @@ def load(
         typer.Option(
             "--type",
             "-t",
-            help="Source type: website, sitemap (auto-detects if not specified).",
+            help="Source type: website, sitemap, discovery (auto-detects if not specified).",
         ),
     ] = None,
     index: Annotated[
         str,
-        typer.Option("--index", "-i", help="Target index name."),
+        typer.Option("--index", "-i", help="Target index name (use tenant-specific name for multi-tenancy)."),
     ] = "knowledge",
-    account_id: Annotated[
-        Optional[str],
-        typer.Option("--account-id", "-a", help="Multi-tenant account ID."),
-    ] = None,
     collection_id: Annotated[
         Optional[str],
         typer.Option("--collection-id", "-c", help="Collection grouping ID."),
@@ -186,16 +187,50 @@ def load(
         bool,
         typer.Option("--verbose", "-V", help="Show per-document progress."),
     ] = False,
+    discovery: Annotated[
+        bool,
+        typer.Option(
+            "--discovery",
+            "-D",
+            help="Use discovery loader to crawl and discover all URLs from the website.",
+        ),
+    ] = False,
+    max_depth: Annotated[
+        int,
+        typer.Option("--max-depth", help="Maximum crawl depth for discovery (default: 3)."),
+    ] = 3,
+    max_pages: Annotated[
+        int,
+        typer.Option("--max-pages", help="Maximum pages to discover (default: 100)."),
+    ] = 100,
+    same_domain: Annotated[
+        bool,
+        typer.Option(
+            "--same-domain/--any-domain",
+            help="Only crawl URLs on the same domain (default: same domain only).",
+        ),
+    ] = True,
 ) -> None:
     """Load and index content from URLs or sitemaps.
 
     Fetches content, chunks it for optimal embedding, and indexes
     into OpenSearch with automatic embedding generation.
 
+    [bold]Multi-tenancy:[/bold]
+    Use --index with tenant-specific index names for isolation
+    (e.g., --index knowledge-{account_id}). Each tenant's data
+    is stored in a separate index for complete isolation.
+
+    [bold]Discovery Mode:[/bold]
+    Use --discovery to crawl and discover all URLs from a website
+    before loading. This is useful for sites without a sitemap.
+
     [bold]Example:[/bold]
         $ gnosisllm-knowledge load https://docs.example.com/intro
         $ gnosisllm-knowledge load https://example.com/sitemap.xml --type sitemap
         $ gnosisllm-knowledge load https://docs.example.com/sitemap.xml --max-urls 500
+        $ gnosisllm-knowledge load https://docs.example.com --discovery --max-depth 5
+        $ gnosisllm-knowledge load https://docs.example.com --index knowledge-tenant-123
     """
     from gnosisllm_knowledge.cli.commands.load import load_command
 
@@ -205,7 +240,6 @@ def load(
             source=source,
             source_type=source_type,
             index_name=index,
-            account_id=account_id,
             collection_id=collection_id,
             source_id=source_id,
             batch_size=batch_size,
@@ -213,6 +247,10 @@ def load(
             force=force,
             dry_run=dry_run,
             verbose=verbose,
+            discovery=discovery,
+            max_depth=max_depth,
+            max_pages=max_pages,
+            same_domain=same_domain,
         )
     )
 
@@ -238,7 +276,7 @@ def search(
     ] = "hybrid",
     index: Annotated[
         str,
-        typer.Option("--index", "-i", help="Index to search."),
+        typer.Option("--index", "-i", help="Index to search (use tenant-specific name for multi-tenancy)."),
     ] = "knowledge",
     limit: Annotated[
         int,
@@ -248,10 +286,6 @@ def search(
         int,
         typer.Option("--offset", "-o", help="Pagination offset."),
     ] = 0,
-    account_id: Annotated[
-        Optional[str],
-        typer.Option("--account-id", "-a", help="Filter by account ID."),
-    ] = None,
     collection_ids: Annotated[
         Optional[str],
         typer.Option("--collection-ids", "-c", help="Filter by collection IDs (comma-separated)."),
@@ -289,10 +323,16 @@ def search(
     - [cyan]hybrid[/cyan]: Combined semantic + keyword (default, best results)
     - [cyan]agentic[/cyan]: AI-powered search with reasoning
 
+    [bold]Multi-tenancy:[/bold]
+    Use --index with tenant-specific index names for isolation
+    (e.g., --index knowledge-{account_id}). Each tenant's data
+    is stored in a separate index for complete isolation.
+
     [bold]Example:[/bold]
         $ gnosisllm-knowledge search "how to configure auth"
         $ gnosisllm-knowledge search "API reference" --mode semantic --limit 10
         $ gnosisllm-knowledge search --interactive
+        $ gnosisllm-knowledge search "query" --index knowledge-tenant-123
     """
     from gnosisllm_knowledge.cli.commands.search import search_command
 
@@ -304,7 +344,6 @@ def search(
             index_name=index,
             limit=limit,
             offset=offset,
-            account_id=account_id,
             collection_ids=collection_ids,
             source_ids=source_ids,
             min_score=min_score,
@@ -451,7 +490,7 @@ def agentic_setup(
 def agentic_chat(
     index: Annotated[
         str,
-        typer.Option("--index", "-i", help="Index to search."),
+        typer.Option("--index", "-i", help="Index to search (use tenant-specific name for multi-tenancy)."),
     ] = "knowledge",
     agent_type: Annotated[
         str,
@@ -461,10 +500,6 @@ def agentic_chat(
             help="Agent type: flow or conversational (default).",
         ),
     ] = "conversational",
-    account_id: Annotated[
-        Optional[str],
-        typer.Option("--account-id", "-a", help="Filter by account ID."),
-    ] = None,
     collection_ids: Annotated[
         Optional[str],
         typer.Option("--collection-ids", "-c", help="Filter by collection IDs (comma-separated)."),
@@ -479,10 +514,15 @@ def agentic_chat(
     Start a conversation with the AI-powered knowledge assistant.
     The agent remembers context for multi-turn dialogue.
 
+    [bold]Multi-tenancy:[/bold]
+    Use --index with tenant-specific index names for isolation
+    (e.g., --index knowledge-{account_id}).
+
     [bold]Example:[/bold]
         $ gnosisllm-knowledge agentic chat
         $ gnosisllm-knowledge agentic chat --type flow
         $ gnosisllm-knowledge agentic chat --verbose
+        $ gnosisllm-knowledge agentic chat --index knowledge-tenant-123
     """
     from gnosisllm_knowledge.cli.commands.agentic import agentic_chat_command
 
@@ -491,7 +531,6 @@ def agentic_chat(
             display=display,
             index_name=index,
             agent_type=agent_type,
-            account_id=account_id,
             collection_ids=collection_ids,
             verbose=verbose,
         )

gnosisllm_knowledge/cli/commands/agentic.py

@@ -4,6 +4,10 @@ Commands:
 - setup: Configure agents in OpenSearch
 - chat: Interactive agentic chat session
 - status: Show agent configuration status
+
+Note:
+    This library is tenant-agnostic. Multi-tenancy is achieved through index
+    isolation - each tenant should use a separate index (e.g., "knowledge-{account_id}").
 """
 
 from __future__ import annotations
@@ -202,17 +206,19 @@ async def agentic_chat_command(
     display: RichDisplayService,
     index_name: str = "knowledge",
     agent_type: str = "conversational",
-    account_id: str | None = None,
     collection_ids: str | None = None,
     verbose: bool = False,
 ) -> None:
     """Interactive agentic chat session.
 
+    Note:
+        Multi-tenancy is achieved through index isolation. Use tenant-specific
+        index names instead (e.g., --index knowledge-tenant-123).
+
     Args:
         display: Display service for output.
-        index_name: Index to search.
+        index_name: Index to search (use tenant-specific name for isolation).
         agent_type: Agent type ('flow' or 'conversational').
-        account_id: Filter by account ID.
         collection_ids: Filter by collection IDs (comma-separated).
         verbose: Show reasoning steps.
     """
@@ -242,7 +248,6 @@ async def agentic_chat_command(
         if agent_type == "conversational":
             return await searcher.create_conversation(
                 name="CLI Chat Session",
-                account_id=account_id,
             )
         return None
 
@@ -291,7 +296,6 @@ async def agentic_chat_command(
                     agent_type=AgentType.CONVERSATIONAL if agent_type == "conversational" else AgentType.FLOW,
                     conversation_id=conversation_id,
                     collection_ids=collection_list,
-                    account_id=account_id,
                     include_reasoning=verbose,
                 )
 
@@ -395,7 +399,6 @@ async def agentic_search_command(
     query: str,
     index_name: str = "knowledge",
     agent_type: str = "flow",
-    account_id: str | None = None,
     collection_ids: str | None = None,
     source_ids: str | None = None,
     limit: int = 5,
@@ -404,12 +407,15 @@ async def agentic_search_command(
 ) -> dict[str, Any] | None:
     """Execute agentic search.
 
+    Note:
+        Multi-tenancy is achieved through index isolation. Use tenant-specific
+        index names instead (e.g., --index knowledge-tenant-123).
+
     Args:
         display: Display service for output.
         query: Search query text.
-        index_name: Index to search.
+        index_name: Index to search (use tenant-specific name for isolation).
         agent_type: Agent type ('flow' or 'conversational').
-        account_id: Filter by account ID.
         collection_ids: Filter by collection IDs (comma-separated).
         source_ids: Filter by source IDs (comma-separated).
         limit: Maximum source documents to retrieve.
@@ -447,12 +453,12 @@ async def agentic_search_command(
             )
 
         # Build query
+        # Note: account_id is deprecated and ignored - use index isolation instead
         agentic_query = AgenticSearchQuery(
             text=query,
             agent_type=AgentType.CONVERSATIONAL if agent_type == "conversational" else AgentType.FLOW,
             collection_ids=collection_list,
             source_ids=source_list,
-            account_id=account_id,
             limit=limit,
             include_reasoning=verbose,
         )