okb 1.0.0__py3-none-any.whl → 1.1.0__py3-none-any.whl

okb/config.py

@@ -53,7 +53,7 @@ class DatabaseConfig:
 
     name: str
     url: str
-    managed: bool = True  # Whether lkb manages this (Docker) or external
+    managed: bool = True  # Whether okb manages this (Docker) or external
     default: bool = False
     description: str | None = None  # Human-readable description for LLM context
     topics: list[str] | None = None  # Topic keywords to help LLM route queries
@@ -259,6 +259,7 @@ DEFAULTS = {
             "yarn.lock",
             "uv.lock",
             "Cargo.lock",
+            "poetry.lock",
             "*.pyc",
             "*.pyo",
             "*.tmp",
@@ -281,7 +282,7 @@ DEFAULTS = {
     },
     "llm": {
         # LLM provider configuration
-        # provider: None = disabled, "claude" = Anthropic API
+        # provider: None = disabled, "claude" = Anthropic API, "modal" = Modal GPU
         "provider": None,
         "model": "claude-haiku-4-5-20251001",
         "timeout": 30,
@@ -289,6 +290,38 @@ DEFAULTS = {
         # Bedrock settings (when use_bedrock is True)
         "use_bedrock": False,
         "aws_region": "us-west-2",
+        # Modal settings (when provider is "modal")
+        "modal_gpu": "L4",  # GPU type: T4, L4, A10G, A100, etc.
+    },
+    "enrichment": {
+        # LLM-based document enrichment
+        "enabled": True,
+        "version": 1,  # Increment to force re-enrichment
+        # What to extract
+        "extract_todos": True,
+        "extract_entities": True,
+        # Auto-create behavior
+        "auto_create_todos": True,      # TODOs created immediately
+        "auto_create_entities": False,  # Entities go to pending_entities table
+        # Confidence thresholds
+        "min_confidence_todo": 0.7,
+        "min_confidence_entity": 0.8,
+        # Auto-enrich during ingest (per source type)
+        "auto_enrich": {
+            "markdown": True,
+            "org": True,
+            "text": True,
+            "code": False,      # Skip code files
+            "web": False,       # Skip web pages
+            "todoist-task": False,  # Already structured
+        },
+        # Entity consolidation settings
+        "consolidation": {
+            "cross_doc_min_mentions": 3,       # Min docs for cross-doc detection
+            "embedding_similarity_threshold": 0.85,  # For duplicate detection
+            "auto_merge_threshold": 0.95,      # Auto-approve above this
+            "min_cluster_size": 3,             # Min entities per cluster
+        },
     },
 }
 
@@ -349,12 +382,30 @@ class Config:
     llm_cache_responses: bool = True
     llm_use_bedrock: bool = False
     llm_aws_region: str = "us-west-2"
+    llm_modal_gpu: str = "L4"
+
+    # Enrichment settings (loaded from config in __post_init__)
+    enrichment_enabled: bool = True
+    enrichment_version: int = 1
+    enrichment_extract_todos: bool = True
+    enrichment_extract_entities: bool = True
+    enrichment_auto_create_todos: bool = True
+    enrichment_auto_create_entities: bool = False
+    enrichment_min_confidence_todo: float = 0.7
+    enrichment_min_confidence_entity: float = 0.8
+    enrichment_auto_enrich: dict[str, bool] = field(default_factory=dict)
+
+    # Consolidation settings (loaded from config in __post_init__)
+    consolidation_cross_doc_min_mentions: int = 3
+    consolidation_embedding_similarity_threshold: float = 0.85
+    consolidation_auto_merge_threshold: float = 0.95
+    consolidation_min_cluster_size: int = 3
 
     def __post_init__(self):
         """Load configuration from file and environment."""
         file_config = load_config_file()
 
-        # Load and merge local config overlay (.lkbconf.yaml)
+        # Load and merge local config overlay (.okbconf.yaml)
         local_path = find_local_config()
         local_default_db: str | None = None
         if local_path:
@@ -417,7 +468,7 @@ class Config:
         else:
             # Legacy: single database_url (env > file > default)
             legacy_url = os.environ.get(
-                "KB_DATABASE_URL",
+                "OKB_DATABASE_URL",
                 file_config.get("database_url", DEFAULTS["databases"]["default"]["url"]),
             )
             self.databases["default"] = DatabaseConfig(
@@ -535,6 +586,55 @@ class Config:
         )
         self.llm_use_bedrock = llm_cfg.get("use_bedrock", DEFAULTS["llm"]["use_bedrock"])
         self.llm_aws_region = llm_cfg.get("aws_region", DEFAULTS["llm"]["aws_region"])
+        self.llm_modal_gpu = os.environ.get(
+            "OKB_MODAL_GPU",
+            llm_cfg.get("modal_gpu", DEFAULTS["llm"]["modal_gpu"]),
+        )
+
+        # Enrichment settings
+        enrich_cfg = file_config.get("enrichment", {})
+        self.enrichment_enabled = enrich_cfg.get("enabled", DEFAULTS["enrichment"]["enabled"])
+        self.enrichment_version = enrich_cfg.get("version", DEFAULTS["enrichment"]["version"])
+        self.enrichment_extract_todos = enrich_cfg.get(
+            "extract_todos", DEFAULTS["enrichment"]["extract_todos"]
+        )
+        self.enrichment_extract_entities = enrich_cfg.get(
+            "extract_entities", DEFAULTS["enrichment"]["extract_entities"]
+        )
+        self.enrichment_auto_create_todos = enrich_cfg.get(
+            "auto_create_todos", DEFAULTS["enrichment"]["auto_create_todos"]
+        )
+        self.enrichment_auto_create_entities = enrich_cfg.get(
+            "auto_create_entities", DEFAULTS["enrichment"]["auto_create_entities"]
+        )
+        self.enrichment_min_confidence_todo = enrich_cfg.get(
+            "min_confidence_todo", DEFAULTS["enrichment"]["min_confidence_todo"]
+        )
+        self.enrichment_min_confidence_entity = enrich_cfg.get(
+            "min_confidence_entity", DEFAULTS["enrichment"]["min_confidence_entity"]
+        )
+        self.enrichment_auto_enrich = enrich_cfg.get(
+            "auto_enrich", DEFAULTS["enrichment"]["auto_enrich"]
+        )
+
+        # Consolidation settings
+        consolidation_cfg = enrich_cfg.get("consolidation", {})
+        self.consolidation_cross_doc_min_mentions = consolidation_cfg.get(
+            "cross_doc_min_mentions",
+            DEFAULTS["enrichment"]["consolidation"]["cross_doc_min_mentions"],
+        )
+        self.consolidation_embedding_similarity_threshold = consolidation_cfg.get(
+            "embedding_similarity_threshold",
+            DEFAULTS["enrichment"]["consolidation"]["embedding_similarity_threshold"],
+        )
+        self.consolidation_auto_merge_threshold = consolidation_cfg.get(
+            "auto_merge_threshold",
+            DEFAULTS["enrichment"]["consolidation"]["auto_merge_threshold"],
+        )
+        self.consolidation_min_cluster_size = consolidation_cfg.get(
+            "min_cluster_size",
+            DEFAULTS["enrichment"]["consolidation"]["min_cluster_size"],
+        )
 
     def get_database(self, name: str | None = None) -> DatabaseConfig:
         """Get database config by name, or default if None."""
@@ -648,6 +748,24 @@ class Config:
                 "cache_responses": self.llm_cache_responses,
                 "use_bedrock": self.llm_use_bedrock,
                 "aws_region": self.llm_aws_region,
+                "modal_gpu": self.llm_modal_gpu,
+            },
+            "enrichment": {
+                "enabled": self.enrichment_enabled,
+                "version": self.enrichment_version,
+                "extract_todos": self.enrichment_extract_todos,
+                "extract_entities": self.enrichment_extract_entities,
+                "auto_create_todos": self.enrichment_auto_create_todos,
+                "auto_create_entities": self.enrichment_auto_create_entities,
+                "min_confidence_todo": self.enrichment_min_confidence_todo,
+                "min_confidence_entity": self.enrichment_min_confidence_entity,
+                "auto_enrich": self.enrichment_auto_enrich,
+                "consolidation": {
+                    "cross_doc_min_mentions": self.consolidation_cross_doc_min_mentions,
+                    "embedding_similarity_threshold": self.consolidation_embedding_similarity_threshold,
+                    "auto_merge_threshold": self.consolidation_auto_merge_threshold,
+                    "min_cluster_size": self.consolidation_min_cluster_size,
+                },
             },
         }
 

okb/http_server.py

@@ -37,9 +37,15 @@ READ_ONLY_TOOLS = frozenset(
         "get_document",
         "list_sources",
         "list_projects",
+        "list_documents_by_project",
         "recent_documents",
         "get_actionable_items",
         "get_database_info",
+        "list_sync_sources",
+        "list_pending_entities",
+        "list_pending_merges",
+        "get_topic_clusters",
+        "get_entity_relationships",
     }
 )
 
@@ -49,6 +55,17 @@ WRITE_TOOLS = frozenset(
         "delete_knowledge",
         "set_database_description",
         "add_todo",
+        "trigger_sync",
+        "trigger_rescan",
+        "enrich_document",
+        "approve_entity",
+        "reject_entity",
+        "analyze_knowledge_base",
+        "find_entity_duplicates",
+        "merge_entities",
+        "approve_merge",
+        "reject_merge",
+        "run_consolidation",
     }
 )
 
@@ -206,6 +223,24 @@ class HTTPMCPServer:
                     content=[TextContent(type="text", text=f"## Projects\n\n{project_list}")]
                 )
 
+            elif name == "list_documents_by_project":
+                project = arguments["project"]
+                limit = arguments.get("limit", 100)
+                docs = kb.list_documents_by_project(project, limit)
+                if not docs:
+                    return CallToolResult(
+                        content=[
+                            TextContent(
+                                type="text", text=f"No documents found for project '{project}'."
+                            )
+                        ]
+                    )
+                output = [f"## Documents in '{project}' ({len(docs)} documents)\n"]
+                for d in docs:
+                    output.append(f"- **{d['title'] or d['source_path']}** ({d['source_type']})")
+                    output.append(f"  - `{d['source_path']}`")
+                return CallToolResult(content=[TextContent(type="text", text="\n".join(output))])
+
             elif name == "recent_documents":
                 from .mcp_server import format_relative_time, get_document_date
 
@@ -263,13 +298,13 @@ class HTTPMCPServer:
                 deleted = kb.delete_knowledge(arguments["source_path"])
                 if deleted:
                     return CallToolResult(
-                        content=[TextContent(type="text", text="Knowledge entry deleted.")]
+                        content=[TextContent(type="text", text="Document deleted.")]
                     )
                 return CallToolResult(
                     content=[
                         TextContent(
                             type="text",
-                            text="Could not delete. Entry not found or not a Claude-saved entry.",
+                            text="Could not delete. Document not found.",
                         )
                     ]
                 )
@@ -349,6 +384,177 @@ class HTTPMCPServer:
                     content=[TextContent(type="text", text="No fields provided to update.")]
                 )
 
+            elif name == "add_todo":
+                result = kb.save_todo(
+                    title=arguments["title"],
+                    content=arguments.get("content"),
+                    due_date=arguments.get("due_date"),
+                    priority=arguments.get("priority"),
+                    project=arguments.get("project"),
+                    tags=arguments.get("tags"),
+                )
+                parts = [
+                    "TODO created:",
+                    f"- Title: {result['title']}",
+                    f"- Path: `{result['source_path']}`",
+                ]
+                if result.get("priority"):
+                    parts.append(f"- Priority: P{result['priority']}")
+                if result.get("due_date"):
+                    parts.append(f"- Due: {result['due_date']}")
+                return CallToolResult(content=[TextContent(type="text", text="\n".join(parts))])
+
+            elif name == "trigger_sync":
+                from .mcp_server import _run_sync
+
+                # Get the db_url from the knowledge base
+                result = _run_sync(
+                    kb.db_url,
+                    sources=arguments.get("sources", []),
+                    sync_all=arguments.get("all", False),
+                    full=arguments.get("full", False),
+                    doc_ids=arguments.get("doc_ids"),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "trigger_rescan":
+                from .mcp_server import _run_rescan
+
+                result = _run_rescan(
+                    kb.db_url,
+                    dry_run=arguments.get("dry_run", False),
+                    delete_missing=arguments.get("delete_missing", False),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "list_sync_sources":
+                from .mcp_server import _list_sync_sources
+
+                token_info = getattr(self.server, "_current_token_info", None)
+                db_name = token_info.database if token_info else config.get_database().name
+                result = _list_sync_sources(kb.db_url, db_name)
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "enrich_document":
+                from .mcp_server import _enrich_document
+
+                result = _enrich_document(
+                    kb.db_url,
+                    source_path=arguments["source_path"],
+                    extract_todos=arguments.get("extract_todos", True),
+                    extract_entities=arguments.get("extract_entities", True),
+                    auto_create_entities=arguments.get("auto_create_entities", False),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "list_pending_entities":
+                from .mcp_server import _list_pending_entities
+
+                result = _list_pending_entities(
+                    kb.db_url,
+                    entity_type=arguments.get("entity_type"),
+                    limit=arguments.get("limit", 20),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "approve_entity":
+                from .mcp_server import _approve_entity
+
+                result = _approve_entity(kb.db_url, arguments["pending_id"])
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "reject_entity":
+                from .mcp_server import _reject_entity
+
+                result = _reject_entity(kb.db_url, arguments["pending_id"])
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "analyze_knowledge_base":
+                from .mcp_server import _analyze_knowledge_base
+
+                result = _analyze_knowledge_base(
+                    kb.db_url,
+                    project=arguments.get("project"),
+                    sample_size=arguments.get("sample_size", 15),
+                    auto_update=arguments.get("auto_update", True),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            # Entity consolidation tools
+            elif name == "find_entity_duplicates":
+                from .mcp_server import _find_entity_duplicates
+
+                result = _find_entity_duplicates(
+                    kb.db_url,
+                    similarity_threshold=arguments.get("similarity_threshold", 0.85),
+                    limit=arguments.get("limit", 50),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "merge_entities":
+                from .mcp_server import _merge_entities
+
+                result = _merge_entities(
+                    kb.db_url,
+                    canonical_path=arguments["canonical_path"],
+                    duplicate_path=arguments["duplicate_path"],
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "list_pending_merges":
+                from .mcp_server import _list_pending_merges
+
+                result = _list_pending_merges(
+                    kb.db_url,
+                    limit=arguments.get("limit", 50),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "approve_merge":
+                from .mcp_server import _approve_merge
+
+                result = _approve_merge(kb.db_url, arguments["merge_id"])
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "reject_merge":
+                from .mcp_server import _reject_merge
+
+                result = _reject_merge(kb.db_url, arguments["merge_id"])
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "get_topic_clusters":
+                from .mcp_server import _get_topic_clusters
+
+                result = _get_topic_clusters(
+                    kb.db_url,
+                    limit=arguments.get("limit", 20),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "get_entity_relationships":
+                from .mcp_server import _get_entity_relationships
+
+                result = _get_entity_relationships(
+                    kb.db_url,
+                    entity_name=arguments.get("entity_name"),
+                    relationship_type=arguments.get("relationship_type"),
+                    limit=arguments.get("limit", 50),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
+            elif name == "run_consolidation":
+                from .mcp_server import _run_consolidation
+
+                result = _run_consolidation(
+                    kb.db_url,
+                    detect_duplicates=arguments.get("detect_duplicates", True),
+                    detect_cross_doc=arguments.get("detect_cross_doc", True),
+                    build_clusters=arguments.get("build_clusters", True),
+                    extract_relationships=arguments.get("extract_relationships", True),
+                    dry_run=arguments.get("dry_run", False),
+                )
+                return CallToolResult(content=[TextContent(type="text", text=result)])
+
             else:
                 return CallToolResult(
                     content=[TextContent(type="text", text=f"Unknown tool: {name}")]