feat: add multi-project support to MCP tools

README.md

@@ -519,7 +519,10 @@ claude mcp add --transport stdio graph-code \
 <!-- SECTION:mcp_tools -->
 | Tool | Description |
 |----|-----------|
-| `index_repository` | Parse and ingest the repository into the Memgraph knowledge graph. This builds a comprehensive graph of functions, classes, dependencies, and relationships. |
+| `list_projects` | List all indexed projects in the knowledge graph database. Returns a list of project names that have been indexed. |
+| `delete_project` | Delete a specific project from the knowledge graph database. This removes all nodes associated with the project while preserving other projects. Use list_projects first to see available projects. |
+| `wipe_database` | WARNING: Completely wipe the entire database, removing ALL indexed projects. This cannot be undone. Use delete_project for removing individual projects. |
+| `index_repository` | Parse and ingest the repository into the Memgraph knowledge graph. This builds a comprehensive graph of functions, classes, dependencies, and relationships. Note: This preserves other projects - only the current project is re-indexed. |
 | `query_code_graph` | Query the codebase knowledge graph using natural language. Ask questions like 'What functions call UserService.create_user?' or 'Show me all classes that implement the Repository interface'. |
 | `get_code_snippet` | Retrieve source code for a function, class, or method by its qualified name. Returns the source code, file path, line numbers, and docstring. |
 | `surgical_replace_code` | Surgically replace an exact code block in a file using diff-match-patch. Only modifies the exact target block, leaving the rest unchanged. |

codebase_rag/constants.py

@@ -169,6 +169,8 @@ class GoogleProviderType(StrEnum):
 KEY_FROM_VAL = "from_val"
 KEY_TO_VAL = "to_val"
 KEY_VERSION_SPEC = "version_spec"
+KEY_PREFIX = "prefix"
+KEY_PROJECT_NAME = "project_name"
 
 ERR_SUBSTR_ALREADY_EXISTS = "already exists"
 ERR_SUBSTR_CONSTRAINT = "constraint"
@@ -2298,6 +2300,9 @@ class CppNodeType(StrEnum):
 
 # (H) MCP tool names
 class MCPToolName(StrEnum):
+    LIST_PROJECTS = "list_projects"
+    DELETE_PROJECT = "delete_project"
+    WIPE_DATABASE = "wipe_database"
     INDEX_REPOSITORY = "index_repository"
     QUERY_CODE_GRAPH = "query_code_graph"
     GET_CODE_SNIPPET = "get_code_snippet"
@@ -2319,6 +2324,7 @@ class MCPSchemaType(StrEnum):
     OBJECT = "object"
     STRING = "string"
     INTEGER = "integer"
+    BOOLEAN = "boolean"
 
 
 # (H) MCP schema fields
@@ -2332,6 +2338,8 @@ class MCPSchemaField(StrEnum):
 
 # (H) MCP parameter names
 class MCPParamName(StrEnum):
+    PROJECT_NAME = "project_name"
+    CONFIRM = "confirm"
     NATURAL_LANGUAGE_QUERY = "natural_language_query"
     QUALIFIED_NAME = "qualified_name"
     FILE_PATH = "file_path"
@@ -2354,10 +2362,15 @@ class MCPParamName(StrEnum):
 
 # (H) MCP response messages
 MCP_INDEX_SUCCESS = "Successfully indexed repository at {path}. Knowledge graph has been updated (previous data cleared)."
+MCP_INDEX_SUCCESS_PROJECT = "Successfully indexed repository at {path}. Project '{project_name}' has been updated."
 MCP_INDEX_ERROR = "Error indexing repository: {error}"
 MCP_WRITE_SUCCESS = "Successfully wrote file: {path}"
 MCP_UNKNOWN_TOOL_ERROR = "Unknown tool: {name}"
 MCP_TOOL_EXEC_ERROR = "Error executing tool '{name}': {error}"
+MCP_PROJECT_DELETED = "Successfully deleted project '{project_name}'."
+MCP_WIPE_CANCELLED = "Database wipe cancelled. Set confirm=true to proceed."
+MCP_WIPE_SUCCESS = "Database completely wiped. All projects have been removed."
+MCP_WIPE_ERROR = "Error wiping database: {error}"
 
 # (H) MCP dict keys and values
 MCP_KEY_RESULTS = "results"

codebase_rag/cypher_queries.py

@@ -1,27 +1,50 @@
 CYPHER_DELETE_ALL = "MATCH (n) DETACH DELETE n;"
 
+CYPHER_LIST_PROJECTS = "MATCH (p:Project) RETURN p.name AS name ORDER BY p.name"
+
+CYPHER_DELETE_PROJECT = """
+MATCH (p:Project {name: $project_name})
+OPTIONAL MATCH (p)-[*]-(connected)
+DETACH DELETE p, connected
+"""
+
 CYPHER_EXAMPLE_DECORATED_FUNCTIONS = """MATCH (n:Function|Method)
 WHERE ANY(d IN n.decorators WHERE toLower(d) IN ['flow', 'task'])
-RETURN n.name AS name, n.qualified_name AS qualified_name, labels(n) AS type"""
+RETURN n.name AS name, n.qualified_name AS qualified_name, labels(n) AS type
+LIMIT 50"""
 
 CYPHER_EXAMPLE_CONTENT_BY_PATH = """MATCH (n)
 WHERE n.path IS NOT NULL AND n.path STARTS WITH 'workflows'
-RETURN n.name AS name, n.path AS path, labels(n) AS type"""
+RETURN n.name AS name, n.path AS path, labels(n) AS type
+LIMIT 50"""
 
 CYPHER_EXAMPLE_KEYWORD_SEARCH = """MATCH (n)
 WHERE toLower(n.name) CONTAINS 'database' OR (n.qualified_name IS NOT NULL AND toLower(n.qualified_name) CONTAINS 'database')
-RETURN n.name AS name, n.qualified_name AS qualified_name, labels(n) AS type"""
+RETURN n.name AS name, n.qualified_name AS qualified_name, labels(n) AS type
+LIMIT 50"""
 
 CYPHER_EXAMPLE_FIND_FILE = """MATCH (f:File) WHERE toLower(f.name) = 'readme.md' AND f.path = 'README.md'
 RETURN f.path as path, f.name as name, labels(f) as type"""
 
-CYPHER_EXAMPLE_README = """MATCH (f:File) WHERE toLower(f.name) CONTAINS 'readme' RETURN f.path AS path, f.name AS name, labels(f) AS type"""
-
-CYPHER_EXAMPLE_PYTHON_FILES = """MATCH (f:File) WHERE f.extension = '.py' RETURN f.path AS path, f.name AS name, labels(f) AS type"""
-
-CYPHER_EXAMPLE_TASKS = """MATCH (n:Function|Method) WHERE 'task' IN n.decorators RETURN n.qualified_name AS qualified_name, n.name AS name, labels(n) AS type"""
-
-CYPHER_EXAMPLE_FILES_IN_FOLDER = """MATCH (f:File) WHERE f.path STARTS WITH 'services' RETURN f.path AS path, f.name AS name, labels(f) AS type"""
+CYPHER_EXAMPLE_README = """MATCH (f:File)
+WHERE toLower(f.name) CONTAINS 'readme'
+RETURN f.path AS path, f.name AS name, labels(f) AS type
+LIMIT 50"""
+
+CYPHER_EXAMPLE_PYTHON_FILES = """MATCH (f:File)
+WHERE f.extension = '.py'
+RETURN f.path AS path, f.name AS name, labels(f) AS type
+LIMIT 50"""
+
+CYPHER_EXAMPLE_TASKS = """MATCH (n:Function|Method)
+WHERE 'task' IN n.decorators
+RETURN n.qualified_name AS qualified_name, n.name AS name, labels(n) AS type
+LIMIT 50"""
+
+CYPHER_EXAMPLE_FILES_IN_FOLDER = """MATCH (f:File)
+WHERE f.path STARTS WITH 'services'
+RETURN f.path AS path, f.name AS name, labels(f) AS type
+LIMIT 50"""
 
 CYPHER_EXAMPLE_LIMIT_ONE = """MATCH (f:File) RETURN f.path as path, f.name as name, labels(f) as type LIMIT 1"""
 

codebase_rag/graph_updater.py

@@ -234,7 +234,7 @@ def __init__(
         self.repo_path = repo_path
         self.parsers = parsers
         self.queries = queries
-        self.project_name = repo_path.name
+        self.project_name = repo_path.resolve().name
         self.simple_name_lookup: SimpleNameLookup = defaultdict(set)
         self.function_registry = FunctionRegistryTrie(
             simple_name_lookup=self.simple_name_lookup

codebase_rag/logs.py

@@ -164,6 +164,8 @@
 MG_BATCH_PARAMS_TRUNCATED = "    Params (first 10 of {count}): {params}..."
 MG_CLEANING_DB = "--- Cleaning database... ---"
 MG_DB_CLEANED = "--- Database cleaned. ---"
+MG_DELETING_PROJECT = "--- Deleting project: {project_name} ---"
+MG_PROJECT_DELETED = "--- Project {project_name} deleted. ---"
 MG_ENSURING_CONSTRAINTS = "Ensuring constraints..."
 MG_CONSTRAINTS_DONE = "Constraints checked/created."
 MG_NODE_BUFFER_FLUSH = (
@@ -568,7 +570,14 @@
 MCP_INDEXING_REPO = "[MCP] Indexing repository at: {path}"
 MCP_CLEARING_DB = "[MCP] Clearing existing database to avoid conflicts..."
 MCP_DB_CLEARED = "[MCP] Database cleared. Starting fresh indexing..."
+MCP_CLEARING_PROJECT = "[MCP] Clearing existing data for project '{project_name}'..."
 MCP_ERROR_INDEXING = "[MCP] Error indexing repository: {error}"
+MCP_LISTING_PROJECTS = "[MCP] Listing all projects..."
+MCP_ERROR_LIST_PROJECTS = "[MCP] Error listing projects: {error}"
+MCP_DELETING_PROJECT = "[MCP] Deleting project: {project_name}"
+MCP_ERROR_DELETE_PROJECT = "[MCP] Error deleting project: {error}"
+MCP_WIPING_DATABASE = "[MCP] Wiping entire database!"
+MCP_ERROR_WIPE = "[MCP] Error wiping database: {error}"
 MCP_QUERY_CODE_GRAPH = "[MCP] query_code_graph: {query}"
 MCP_QUERY_RESULTS = "[MCP] Query returned {count} results"
 MCP_ERROR_QUERY = "[MCP] Error querying code graph: {error}"

codebase_rag/mcp/tools.py

@@ -23,6 +23,12 @@
 from codebase_rag.tools.file_writer import FileWriter, create_file_writer_tool
 from codebase_rag.types_defs import (
     CodeSnippetResultDict,
+    DeleteProjectErrorResult,
+    DeleteProjectResult,
+    DeleteProjectSuccessResult,
+    ListProjectsErrorResult,
+    ListProjectsResult,
+    ListProjectsSuccessResult,
     MCPHandlerType,
     MCPInputSchema,
     MCPInputSchemaProperty,
@@ -62,6 +68,49 @@ def __init__(
         )
 
         self._tools: dict[str, ToolMetadata] = {
+            cs.MCPToolName.LIST_PROJECTS: ToolMetadata(
+                name=cs.MCPToolName.LIST_PROJECTS,
+                description=td.MCP_TOOLS[cs.MCPToolName.LIST_PROJECTS],
+                input_schema=MCPInputSchema(
+                    type=cs.MCPSchemaType.OBJECT,
+                    properties={},
+                    required=[],
+                ),
+                handler=self.list_projects,
+                returns_json=True,
+            ),
+            cs.MCPToolName.DELETE_PROJECT: ToolMetadata(
+                name=cs.MCPToolName.DELETE_PROJECT,
+                description=td.MCP_TOOLS[cs.MCPToolName.DELETE_PROJECT],
+                input_schema=MCPInputSchema(
+                    type=cs.MCPSchemaType.OBJECT,
+                    properties={
+                        cs.MCPParamName.PROJECT_NAME: MCPInputSchemaProperty(
+                            type=cs.MCPSchemaType.STRING,
+                            description=td.MCP_PARAM_PROJECT_NAME,
+                        )
+                    },
+                    required=[cs.MCPParamName.PROJECT_NAME],
+                ),
+                handler=self.delete_project,
+                returns_json=True,
+            ),
+            cs.MCPToolName.WIPE_DATABASE: ToolMetadata(
+                name=cs.MCPToolName.WIPE_DATABASE,
+                description=td.MCP_TOOLS[cs.MCPToolName.WIPE_DATABASE],
+                input_schema=MCPInputSchema(
+                    type=cs.MCPSchemaType.OBJECT,
+                    properties={
+                        cs.MCPParamName.CONFIRM: MCPInputSchemaProperty(
+                            type=cs.MCPSchemaType.BOOLEAN,
+                            description=td.MCP_PARAM_CONFIRM,
+                        )
+                    },
+                    required=[cs.MCPParamName.CONFIRM],
+                ),
+                handler=self.wipe_database,
+                returns_json=False,
+            ),
             cs.MCPToolName.INDEX_REPOSITORY: ToolMetadata(
                 name=cs.MCPToolName.INDEX_REPOSITORY,
                 description=td.MCP_TOOLS[cs.MCPToolName.INDEX_REPOSITORY],
@@ -199,13 +248,53 @@ def __init__(
             ),
         }
 
-    async def index_repository(self) -> str:
-        logger.info(lg.MCP_INDEXING_REPO.format(path=self.project_root))
+    async def list_projects(self) -> ListProjectsResult:
+        logger.info(lg.MCP_LISTING_PROJECTS)
+        try:
+            projects = self.ingestor.list_projects()
+            return ListProjectsSuccessResult(projects=projects, count=len(projects))
+        except Exception as e:
+            logger.error(lg.MCP_ERROR_LIST_PROJECTS.format(error=e))
+            return ListProjectsErrorResult(error=str(e), projects=[], count=0)
 
+    async def delete_project(self, project_name: str) -> DeleteProjectResult:
+        logger.info(lg.MCP_DELETING_PROJECT.format(project_name=project_name))
+        try:
+            projects = self.ingestor.list_projects()
+            if project_name not in projects:
+                return DeleteProjectErrorResult(
+                    success=False,
+                    error=te.MCP_PROJECT_NOT_FOUND.format(
+                        project_name=project_name, projects=projects
+                    ),
+                )
+            self.ingestor.delete_project(project_name)
+            return DeleteProjectSuccessResult(
+                success=True,
+                project=project_name,
+                message=cs.MCP_PROJECT_DELETED.format(project_name=project_name),
+            )
+        except Exception as e:
+            logger.error(lg.MCP_ERROR_DELETE_PROJECT.format(error=e))
+            return DeleteProjectErrorResult(success=False, error=str(e))
+
+    async def wipe_database(self, confirm: bool) -> str:
+        if not confirm:
+            return cs.MCP_WIPE_CANCELLED
+        logger.warning(lg.MCP_WIPING_DATABASE)
         try:
-            logger.info(lg.MCP_CLEARING_DB)
             self.ingestor.clean_database()
-            logger.info(lg.MCP_DB_CLEARED)
+            return cs.MCP_WIPE_SUCCESS
+        except Exception as e:
+            logger.error(lg.MCP_ERROR_WIPE.format(error=e))
+            return cs.MCP_WIPE_ERROR.format(error=e)
+
+    async def index_repository(self) -> str:
+        logger.info(lg.MCP_INDEXING_REPO.format(path=self.project_root))
+        project_name = Path(self.project_root).resolve().name
+        try:
+            logger.info(lg.MCP_CLEARING_PROJECT.format(project_name=project_name))
+            self.ingestor.delete_project(project_name)
 
             updater = GraphUpdater(
                 ingestor=self.ingestor,
@@ -215,7 +304,9 @@ async def index_repository(self) -> str:
             )
             updater.run()
 
-            return cs.MCP_INDEX_SUCCESS.format(path=self.project_root)
+            return cs.MCP_INDEX_SUCCESS_PROJECT.format(
+                path=self.project_root, project_name=project_name
+            )
         except Exception as e:
             logger.error(lg.MCP_ERROR_INDEXING.format(error=e))
             return cs.MCP_INDEX_ERROR.format(error=e)

codebase_rag/prompts.py

@@ -133,8 +133,20 @@ def build_rag_orchestrator_prompt(tools: list["Tool"]) -> str:
 
 {GRAPH_SCHEMA_AND_RULES}
 
-**3. Query Patterns & Examples**
-Your goal is to return the `name`, `path`, and `qualified_name` of the found nodes.
+**3. Query Optimization Rules**
+
+- **LIMIT Results**: ALWAYS add `LIMIT 50` to queries that list items. This prevents overwhelming responses.
+- **Aggregation Queries**: When asked "how many", "count", or "total", return ONLY the count, not all items:
+  - CORRECT: `MATCH (c:Class) RETURN count(c) AS total`
+  - WRONG: `MATCH (c:Class) RETURN c.name, c.path, count(c) AS total` (returns all items!)
+- **List vs Count**: If asked to "list" or "show", return items with LIMIT. If asked to "count" or "how many", return only the count.
+
+**4. Query Patterns & Examples**
+When listing items, return the `name`, `path`, and `qualified_name` with a LIMIT.
+
+**Pattern: Counting Items**
+cypher// "How many classes are there?" or "Count all functions"
+MATCH (c:Class) RETURN count(c) AS total
 
 **Pattern: Finding Decorated Functions/Methods (e.g., Workflows, Tasks)**
 cypher// "Find all prefect flows" or "what are the workflows?" or "show me the tasks"
@@ -172,9 +184,19 @@ def build_rag_orchestrator_prompt(tools: list["Tool"]) -> str:
     - For code nodes (`Class`, `Function`, etc.), return `n.qualified_name AS qualified_name`.
 4.  **KEEP IT SIMPLE**: Do not try to be clever. A simple query that returns a few relevant nodes is better than a complex one that fails.
 5.  **CLAUSE ORDER**: You MUST follow the standard Cypher clause order: `MATCH`, `WHERE`, `RETURN`, `LIMIT`.
+6.  **ALWAYS ADD LIMIT**: For queries that list items, ALWAYS add `LIMIT 50` to prevent overwhelming responses.
+7.  **AGGREGATION QUERIES**: When asked "how many" or "count", return ONLY the count:
+    - CORRECT: `MATCH (c:Class) RETURN count(c) AS total`
+    - WRONG: `MATCH (c:Class) RETURN c.name, count(c) AS total` (returns all items!)
 
 **Examples:**
 
+*   **Natural Language:** "How many classes are there?"
+*   **Cypher Query:**
+    ```cypher
+    MATCH (c:Class) RETURN count(c) AS total
+    ```
+
 *   **Natural Language:** "Find the main README file"
 *   **Cypher Query:**
     ```cypher

codebase_rag/services/graph_service.py

@@ -15,15 +15,19 @@
     ERR_SUBSTR_CONSTRAINT,
     KEY_CREATED,
     KEY_FROM_VAL,
+    KEY_NAME,
+    KEY_PROJECT_NAME,
     KEY_PROPS,
     KEY_TO_VAL,
     NODE_UNIQUE_CONSTRAINTS,
     REL_TYPE_CALLS,
 )
 from ..cypher_queries import (
     CYPHER_DELETE_ALL,
+    CYPHER_DELETE_PROJECT,
     CYPHER_EXPORT_NODES,
     CYPHER_EXPORT_RELATIONSHIPS,
+    CYPHER_LIST_PROJECTS,
     build_constraint_query,
     build_merge_node_query,
     build_merge_relationship_query,
@@ -164,6 +168,15 @@ def clean_database(self) -> None:
         self._execute_query(CYPHER_DELETE_ALL)
         logger.info(ls.MG_DB_CLEANED)
 
+    def list_projects(self) -> list[str]:
+        result = self.fetch_all(CYPHER_LIST_PROJECTS)
+        return [str(r[KEY_NAME]) for r in result]
+
+    def delete_project(self, project_name: str) -> None:
+        logger.info(ls.MG_DELETING_PROJECT.format(project_name=project_name))
+        self._execute_query(CYPHER_DELETE_PROJECT, {KEY_PROJECT_NAME: project_name})
+        logger.info(ls.MG_PROJECT_DELETED.format(project_name=project_name))
+
     def ensure_constraints(self) -> None:
         logger.info(ls.MG_ENSURING_CONSTRAINTS)
         for label, prop in NODE_UNIQUE_CONSTRAINTS.items():