codebase-retrieval-context-engine 2.0.0__py3-none-any.whl → 2.0.2__py3-none-any.whl

{codebase_retrieval_context_engine-2.0.0.dist-info → codebase_retrieval_context_engine-2.0.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codebase-retrieval-context-engine
-Version: 2.0.0
+Version: 2.0.2
 Summary: Code retrieval engine — hybrid embedding + graph search for LLM context injection.
 Project-URL: Homepage, https://github.com/nullmastermind/local-context-engine
 Project-URL: Repository, https://github.com/nullmastermind/local-context-engine
@@ -17,6 +17,7 @@ Classifier: Topic :: Software Development :: Libraries
 Requires-Python: >=3.11
 Requires-Dist: mcp>=1.1.2
 Requires-Dist: numpy>=2.0
+Requires-Dist: pathspec>=0.11
 Requires-Dist: pydantic>=2.0
 Requires-Dist: python-dotenv>=1.0
 Requires-Dist: rich>=13.0
@@ -40,9 +41,6 @@ Requires-Dist: anthropic[vertex]>=0.25; extra == 'gcp'
 Requires-Dist: google-cloud-aiplatform>=1.38; extra == 'gcp'
 Provides-Extra: google
 Requires-Dist: google-genai>=2.7.0; extra == 'google'
-Provides-Extra: local
-Requires-Dist: sentence-transformers>=3.0; extra == 'local'
-Requires-Dist: transformers<5.0.0; extra == 'local'
 Provides-Extra: openai
 Requires-Dist: openai>=1.0; extra == 'openai'
 Provides-Extra: treesitter

{codebase_retrieval_context_engine-2.0.0.dist-info → codebase_retrieval_context_engine-2.0.2.dist-info}/RECORD

@@ -1,4 +1,4 @@
-corbell/__init__.py,sha256=ECnnvynYrlq_niQjMhnC171R8q2axnwhEGqVRoWu50E,124
+corbell/__init__.py,sha256=DK8C29me67FSOnq2v_CAPc0COnXW4plMGTNHfZvmX5Y,124
 corbell/cli/__init__.py,sha256=5-MP6JIWgp4nDLNIhqP6Gtx97GESaIYg3NGxtRGaMv0,28
 corbell/cli/main.py,sha256=anYpXiyQD6_1wMS0Dtef6Rxtxd0NEFe7HHnerHxf3J4,1835
 corbell/cli/commands/__init__.py,sha256=0mAOs3RWC7XMZnGRN677hjPCHHQKDq9ASjIr_GQM3js,37
@@ -8,39 +8,39 @@ corbell/core/__init__.py,sha256=VS9PnhHr4NXYlWs1TLCyllnVCNsiwVZ1Xj-AOBhZpAU,29
 corbell/core/constants.py,sha256=HTGYpShlp9pP2_a4WngHtTujUQfHcypFAYoaczmkBdQ,1061
 corbell/core/gitignore.py,sha256=VS7_s6NwZWQAwgLiaRzPHdBRIj86XdnPm_P_x_e0hvI,2266
 corbell/core/llm_client.py,sha256=2MDwe6kr_EyY3DFv3fNO91WCig8ER021ogzdLGH3IN8,26219
-corbell/core/workspace.py,sha256=v9H56MxhI2iHPr3P9fU5DnIiniGehozWScfcd6TG2gQ,14189
+corbell/core/workspace.py,sha256=NsfByxnqTbPeflXLBqXAkqVaQCQ9Qs9maUmxp2Y6n1k,14024
 corbell/core/embeddings/__init__.py,sha256=RCekvfNkFuMGEDLnls78i3znR84cTdnj4KJ_PeQrMNg,213
 corbell/core/embeddings/base.py,sha256=udPW4XmcPhCpNQA6n8KqMcu2JXvVNv1JjdRJmFq5ZRA,2175
 corbell/core/embeddings/extractor.py,sha256=hOolMX6JX3sVBf062h2zUQpr9SVt81S0hzhNCeJoV1I,7180
 corbell/core/embeddings/factory.py,sha256=Lonjbk8Lsxykz-2ZEgFCWoH9zZ005Qm4dXVdA6P4qJY,1817
-corbell/core/embeddings/model.py,sha256=5guf3qpqOaC7fSp-OZlQjBUQt7jOf_3eE3w6X6yNY2I,15224
+corbell/core/embeddings/model.py,sha256=sKFjUYJ8-COth1CXjgX9Bn_oPcf1OSbbq04oSywMDSo,14128
 corbell/core/embeddings/search_cache.py,sha256=FHzO3mu4m4MJGy2jOFwb9GCEypcT11CcVrLts4Ib0ho,3351
 corbell/core/embeddings/sqlite_store.py,sha256=8rv89WOMqMm-JhJO36-FdRiC68Ija3TwHkrmRrPr1os,10158
 corbell/core/graph/__init__.py,sha256=VaxDKeXMgMEBBMC0dglwj68A_aNYRI5O8VM6oMC1GIM,29
-corbell/core/graph/builder.py,sha256=8BtUVvh_3Sq_XQiYf3yZifRygTrc-_hE770luqwZ3e0,30423
-corbell/core/graph/method_graph.py,sha256=XJpojhBVXon2RNMIaOqE3NgRjUdP0IAnRQuKYGhO5js,49753
+corbell/core/graph/builder.py,sha256=_TjcKfOKObeJ3ScCMLZNHhtzmBYs1VtJEEp3UJLfoO0,32118
+corbell/core/graph/method_graph.py,sha256=x6X91Dz3DzNAuzld2f7ORkODt3qC5L1Fzg1bdAcIhK4,50851
 corbell/core/graph/schema.py,sha256=swy1VZZpL88LPEj6zihl5bglQLrGD-ohOYjFeNC31a0,5253
-corbell/core/graph/sqlite_store.py,sha256=D9que8zHV68x_tiFY-q09Ipk-dXR9LWM3hFPQLGNDto,20019
+corbell/core/graph/sqlite_store.py,sha256=B1ObNit7MXbQpst6dpuloTcFAmUim_MoP3PSCATf_4A,21116
 corbell/core/graph/providers/__init__.py,sha256=__ZVe1uwIHSyFh_t-V4MyT5MsM5hooTOrxxkm9Txt7o,268
 corbell/core/graph/providers/aws_patterns.py,sha256=w2iF5qQJcV7S6J64ZYb3IzGPdXjCc37YX5sNnHz8mXY,2818
 corbell/core/graph/providers/azure_patterns.py,sha256=tJ9AQQXW2xYzJ36wNOxTHHhaivaCv3RYEMJUjw8WjeQ,3515
 corbell/core/graph/providers/gcp_patterns.py,sha256=vIofjanvRWGhFftuGdzt9YgTIGZRJz7lLG0abUNjFdA,2789
 corbell/core/indexing/__init__.py,sha256=VczeSHUfKR3YVowGCleFjo2pIpDHfl9kl-OkEl8szow,47
-corbell/core/indexing/builder.py,sha256=l1Hj9Lrze2zuBvwtauDgmjAWWVvKbKXWKMgtv8uA-TQ,21967
+corbell/core/indexing/builder.py,sha256=mxWdHqgAx6akO8vb8-tlshD4zTlmbRuR-TOt-jETDLs,23303
 corbell/core/indexing/lock.py,sha256=uUMelIrtrp6Ww9rTfbl2OvomByc-IJyiHIMnptfA4xI,4743
 corbell/core/indexing/tracker.py,sha256=mbL1M-EeYf6KoIT5qoz7LCHwSHL6UlZNX7mjm4DczR0,8469
 corbell/core/mcp/__init__.py,sha256=DDzfuVbX_GBTM5Nqy34JVgDUMeFd2_5ZcVMVuvjOddU,32
 corbell/core/mcp/server.py,sha256=nTiPQ9yyenL7uhgLCsGwEm7yyoqk1tUPTsZYFAAmPBU,7270
 corbell/core/query/__init__.py,sha256=OCyVRZOyh_eLGhOxR_JYyH6zp8O7qy_-rC3fqGHm7Bc,56
 corbell/core/query/diagnostics.py,sha256=ObQyZWmMVRXEHFYGXBP2-EMBmM8SYr0H6cCi95uFnIk,1406
-corbell/core/query/engine.py,sha256=4PRW3R0Wws7BPAqDD4hrWjpqOq8mx9dCrtABWA-Wmus,11635
+corbell/core/query/engine.py,sha256=cEueZdZQcg_o5HaPaayE4hCGiCvyIxvv0OWnXWD2DzU,11855
 corbell/core/query/enhancer.py,sha256=w5mvm1B8qQZpL6RVhMuhq_rls77hakGSNUyanfkyNEU,3934
 corbell/core/query/formatter.py,sha256=xMr8HE-oxBSEKb514aixY7aoUWGeYoK1w5wnaIlCYEc,2813
 corbell/core/query/graph_expander.py,sha256=Y-yKnr6db-OM2Gh8ukYgVIcUZa6-wfWA-GhdvOwf_yA,9184
 corbell/core/query/merger.py,sha256=fs6PL7X7EweXnSnDRnpzmpaU8JjwJpL0akzm4hSwLJk,6168
 corbell/core/query/reranker.py,sha256=HYckYiUVZ80mbLGHhK4IHxNI7uUqNaztwXLbYgdnoWU,4298
-codebase_retrieval_context_engine-2.0.0.dist-info/METADATA,sha256=SVKzXFoaizz5EyBgOZbmwdtsKI72ZcKgjzCyes_8mJY,17408
-codebase_retrieval_context_engine-2.0.0.dist-info/WHEEL,sha256=mffPy8wBnZQn2VnJUU5jE99KsxaSfiyMHV9Yt0aLVxs,87
-codebase_retrieval_context_engine-2.0.0.dist-info/entry_points.txt,sha256=vFB4a4Qb7Ty182usK8deJXiis0UYnGIUDusw0V3Jya8,115
-codebase_retrieval_context_engine-2.0.0.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
-codebase_retrieval_context_engine-2.0.0.dist-info/RECORD,,
+codebase_retrieval_context_engine-2.0.2.dist-info/METADATA,sha256=20ALXtYeqFP5ZR_j0hsKhZpI1YAxRvLFm2CzM8BRHgQ,17304
+codebase_retrieval_context_engine-2.0.2.dist-info/WHEEL,sha256=mffPy8wBnZQn2VnJUU5jE99KsxaSfiyMHV9Yt0aLVxs,87
+codebase_retrieval_context_engine-2.0.2.dist-info/entry_points.txt,sha256=vFB4a4Qb7Ty182usK8deJXiis0UYnGIUDusw0V3Jya8,115
+codebase_retrieval_context_engine-2.0.2.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
+codebase_retrieval_context_engine-2.0.2.dist-info/RECORD,,

corbell/__init__.py

@@ -2,5 +2,5 @@
 Corbell — Code retrieval engine for LLM context injection.
 """
 
-__version__ = "2.0.0"
+__version__ = "2.0.2"
 __all__ = ["__version__"]

corbell/core/embeddings/model.py

@@ -1,4 +1,4 @@
-"""Embedding model interface + SentenceTransformers implementation."""
+"""Embedding model interface + cloud provider implementations (Google, Voyage)."""
 
 from __future__ import annotations
 
@@ -33,32 +33,6 @@ class EmbeddingModel(ABC):
         ...
 
 
-class SentenceTransformerModel(EmbeddingModel):
-    """Wraps ``sentence-transformers`` with lazy loading.
-
-    Uses ``all-MiniLM-L6-v2`` by default (384-dim, fast, no API key).
-    """
-
-    def __init__(self, model_name: str = "all-MiniLM-L6-v2"):
-        self.model_name = model_name
-        self._model = None  # lazy-loaded
-
-    def _get_model(self):
-        if self._model is None:
-            from sentence_transformers import SentenceTransformer
-            self._model = SentenceTransformer(f"sentence-transformers/{self.model_name}")
-        return self._model
-
-    def encode(self, texts: List[str]) -> List[List[float]]:
-        model = self._get_model()
-        vecs = model.encode(texts, show_progress_bar=False)
-        return [v.tolist() for v in vecs]
-
-    @property
-    def dimension(self) -> int:
-        return self._get_model().get_sentence_embedding_dimension()
-
-
 def _is_voyage_rate_limit_error(e: Exception) -> bool:
     """Return True when a Voyage API error is a 429 rate limit."""
     status = getattr(e, "status_code", None)
@@ -130,9 +104,8 @@ class GoogleEmbeddingModel(EmbeddingModel):
         if not self._api_keys:
             raise ValueError(
                 "GOOGLE_API_KEY is not set. "
-                "Set it in your environment or workspace.yaml:\n"
-                "  export GOOGLE_API_KEY=AIza...\n"
-                "Or use a local embedding model (e.g. all-MiniLM-L6-v2) in storage.model."
+                "Set it in your environment or .env file:\n"
+                "  export GOOGLE_API_KEY=AIza..."
             )
         self._key_index: int = random.randrange(len(self._api_keys))
         # kept for backwards-compat with tests that read _api_key directly
@@ -302,9 +275,8 @@ class VoyageEmbeddingModel(EmbeddingModel):
         if not self._api_keys:
             raise ValueError(
                 "VOYAGE_API_KEY is not set. "
-                "Set it in your environment or workspace.yaml:\n"
-                "  export VOYAGE_API_KEY=pa-...\n"
-                "Or use a local embedding model (e.g. all-MiniLM-L6-v2) in storage.model."
+                "Set it in your environment or .env file:\n"
+                "  export VOYAGE_API_KEY=pa-..."
             )
         self._key_index: int = random.randrange(len(self._api_keys))
         # kept for backwards-compat with tests that read _api_key directly

corbell/core/graph/builder.py

@@ -244,12 +244,16 @@ class ServiceGraphBuilder:
             graph_store: Instance of :class:`~corbell.core.graph.schema.GraphStore`.
         """
         self.store = graph_store
+        self._content_cache: Dict[Path, str] = {}
+        self._pending_nodes: list = []
+        self._pending_edges: list = []
 
     def build_from_workspace(
         self,
         services: List[Dict[str, Any]],
         clear_existing: bool = True,
         method_level: bool = False,
+        file_list: Optional[List[Path]] = None,
     ) -> Dict[str, Any]:
         """Scan all service repos and populate the graph.
 
@@ -258,6 +262,9 @@ class ServiceGraphBuilder:
                 ``language``, ``tags``.
             clear_existing: Clear the store before building.
             method_level: If True, also build method-call edges.
+            file_list: Optional pre-filtered list of Path objects covering all
+                repos. When provided, used instead of calling ``_iter_files``
+                (skips rglob). Falls back to rglob when None.
 
         Returns:
             Summary dict with counts of services, datastores, queues, methods.
@@ -265,6 +272,9 @@ class ServiceGraphBuilder:
         if clear_existing:
             self.store.clear()
 
+        self._pending_nodes = []
+        self._pending_edges = []
+
         discovered: List[Dict] = []
 
         for svc in services:
@@ -277,8 +287,17 @@ class ServiceGraphBuilder:
                 continue
 
             # Gather all relevant files first so we can sniff the service type
-            gitignore_spec = load_gitignore(repo_path)
-            files = list(self._iter_files(repo_path, language, gitignore_spec))
+            if file_list is not None:
+                # Use pre-filtered list — filter to files under this repo_path
+                repo_path_str = str(repo_path)
+                files = [
+                    fp for fp in file_list
+                    if str(fp).startswith(repo_path_str)
+                ]
+            else:
+                gitignore_spec = load_gitignore(repo_path)
+                files = list(self._iter_files(repo_path, language, gitignore_spec))
+
             service_type = self._detect_service_type(files, language)
 
             node = ServiceNode(
@@ -289,7 +308,7 @@ class ServiceGraphBuilder:
                 tags=tags,
                 service_type=service_type,
             )
-            self.store.upsert_node(node)
+            self._pending_nodes.append(node)
             discovered.append(
                 {
                     "id": svc_id,
@@ -300,6 +319,16 @@ class ServiceGraphBuilder:
             )
 
         # Phase 2: deps, HTTP calls
+        # Populate content cache for all discovered files (read once, used across phases)
+        self._content_cache = {}
+        for svc in discovered:
+            for fp in svc["files"]:
+                if fp not in self._content_cache:
+                    try:
+                        self._content_cache[fp] = fp.read_text(encoding="utf-8", errors="ignore")
+                    except Exception:
+                        self._content_cache[fp] = ""
+
         datastore_ids: set = set()
         queue_ids: set = set()
 
@@ -313,6 +342,15 @@ class ServiceGraphBuilder:
             self._detect_http_calls(svc, discovered)
             self._detect_library_deps(svc, all_service_ids)
 
+        # Free cached content
+        self._content_cache = {}
+
+        # Flush all accumulated nodes and edges in two bulk writes
+        self.store.upsert_nodes_batch(self._pending_nodes)
+        self.store.upsert_edges_batch(self._pending_edges)
+        self._pending_nodes = []
+        self._pending_edges = []
+
         # Phase 4: method-level graph
         service_diagnostics: Dict[str, Any] = {}
         if method_level:
@@ -405,6 +443,8 @@ class ServiceGraphBuilder:
         return False
 
     def _read(self, fp: Path) -> str:
+        if fp in self._content_cache:
+            return self._content_cache[fp]
         try:
             return fp.read_text(encoding="utf-8", errors="ignore")
         except Exception:
@@ -486,10 +526,10 @@ class ServiceGraphBuilder:
                     ds_id = f"datastore:{db_type}:{db_name}"
                     if ds_id not in datastore_ids:
                         datastore_ids.add(ds_id)
-                        self.store.upsert_node(DataStoreNode(id=ds_id, kind=db_type, name=db_name))
-                    
+                        self._pending_nodes.append(DataStoreNode(id=ds_id, kind=db_type, name=db_name))
+
                     direction = self._classify_io_direction(content, idx)
-                    self.store.upsert_edge(
+                    self._pending_edges.append(
                         DependencyEdge(
                             source_id=svc_id,
                             target_id=ds_id,
@@ -516,11 +556,11 @@ class ServiceGraphBuilder:
                     q_id = f"queue:{q_type}:{q_name}"
                     if q_id not in queue_ids:
                         queue_ids.add(q_id)
-                        self.store.upsert_node(QueueNode(id=q_id, kind=q_type, name=q_name))
-                    
+                        self._pending_nodes.append(QueueNode(id=q_id, kind=q_type, name=q_name))
+
                     direction = self._classify_io_direction(content, idx)
                     edge_kind = "queue_publish" if direction == "write" else "queue_consume"
-                    self.store.upsert_edge(
+                    self._pending_edges.append(
                         DependencyEdge(
                             source_id=svc_id,
                             target_id=q_id,
@@ -552,7 +592,7 @@ class ServiceGraphBuilder:
                     svc_slug = other_id.replace("-", "").replace("_", "").lower()
                     url_clean = url_host.replace("-", "").replace("_", "").lower()
                     if svc_slug in url_clean:
-                        self.store.upsert_edge(
+                        self._pending_edges.append(
                             DependencyEdge(
                                 source_id=svc_id,
                                 target_id=other_id,
@@ -584,7 +624,7 @@ class ServiceGraphBuilder:
                                     break
                             
                             if mapped_svc_id:
-                                self.store.upsert_edge(
+                                self._pending_edges.append(
                                     DependencyEdge(
                                         source_id=svc_id,
                                         target_id=mapped_svc_id,
@@ -597,7 +637,7 @@ class ServiceGraphBuilder:
                                     )
                                 )
                             else:
-                                self.store.upsert_edge(
+                                self._pending_edges.append(
                                     DependencyEdge(
                                         source_id=svc_id,
                                         target_id="external:env_url",
@@ -630,7 +670,7 @@ class ServiceGraphBuilder:
                         break
                 
                 if mapped_svc_id:
-                    self.store.upsert_edge(
+                    self._pending_edges.append(
                         DependencyEdge(
                             source_id=svc_id,
                             target_id=mapped_svc_id,
@@ -665,7 +705,7 @@ class ServiceGraphBuilder:
                 content = self._read(fp)
                 for exact_name, target_id in exact_to_id.items():
                     if f'"{exact_name}"' in content or f"'{exact_name}'" in content or f" {exact_name}==" in content:
-                        self.store.upsert_edge(
+                        self._pending_edges.append(
                             DependencyEdge(
                                 source_id=svc_id,
                                 target_id=target_id,
@@ -685,7 +725,7 @@ class ServiceGraphBuilder:
                     import_pattern_js = rf"(?:import|require).*{exact_name}"
                     
                     if re.search(import_pattern_py, content) or re.search(import_pattern_js, content):
-                        self.store.upsert_edge(
+                        self._pending_edges.append(
                             DependencyEdge(
                                 source_id=svc_id,
                                 target_id=target_id,

corbell/core/graph/method_graph.py

@@ -297,7 +297,9 @@ class MethodGraphBuilder:
     def __init__(self, graph_store: GraphStore):
         self.store = graph_store
 
-    def build_for_service(self, service_id: str, repo_path: Path) -> Dict[str, Any]:
+    def build_for_service(
+        self, service_id: str, repo_path: Path, file_list: Optional[List[Path]] = None
+    ) -> Dict[str, Any]:
         """Scan *repo_path* and populate method nodes + call edges.
 
         Uses tree-sitter for all supported languages when the grammar packages
@@ -307,6 +309,10 @@ class MethodGraphBuilder:
         Args:
             service_id: Identifier for the owning service.
             repo_path: Root directory of the repository to scan.
+            file_list: Optional pre-filtered list of Path objects to scan.
+                When provided, skips rglob/gitignore/skip_dirs filtering —
+                the caller is responsible for pre-filtering. Falls back to
+                the original rglob behavior when None.
 
         Returns:
             Summary dict with ``methods``, ``calls``, ``files_scanned``, ``ts_available``.
@@ -315,28 +321,41 @@ class MethodGraphBuilder:
         all_calls: List[Dict] = []
         files_scanned = 0
 
-        gitignore_spec = load_gitignore(Path(repo_path))
-
-        for fp in Path(repo_path).rglob("*"):
-            if not fp.is_file():
-                continue
-            # Only skip if the immediate parent directory name is in SKIP_DIRS
-            # (avoids false-positives from matching path segments like 'corbel')
-            rel = fp.relative_to(repo_path)
-            if any(part in _SKIP_DIRS for part in rel.parts):
-                continue
-            if gitignore_spec.match_file(str(rel).replace("\\", "/")):
-                continue
-            lang = _EXT_LANG.get(fp.suffix)
-            if not lang:
-                continue
-            files_scanned += 1
-            result = self._analyze_file(fp, service_id, lang)
-            for m in result["methods"]:
-                all_methods[m["id"]] = m
-            all_calls.extend(result["calls"])
-
-        # Persist method nodes
+        if file_list is not None:
+            # Pre-filtered list from caller — skip all filtering
+            for fp in file_list:
+                lang = _EXT_LANG.get(fp.suffix)
+                if not lang:
+                    continue
+                files_scanned += 1
+                result = self._analyze_file(fp, service_id, lang)
+                for m in result["methods"]:
+                    all_methods[m["id"]] = m
+                all_calls.extend(result["calls"])
+        else:
+            gitignore_spec = load_gitignore(Path(repo_path))
+
+            for fp in Path(repo_path).rglob("*"):
+                if not fp.is_file():
+                    continue
+                # Only skip if the immediate parent directory name is in SKIP_DIRS
+                # (avoids false-positives from matching path segments like 'corbel')
+                rel = fp.relative_to(repo_path)
+                if any(part in _SKIP_DIRS for part in rel.parts):
+                    continue
+                if gitignore_spec.match_file(str(rel).replace("\\", "/")):
+                    continue
+                lang = _EXT_LANG.get(fp.suffix)
+                if not lang:
+                    continue
+                files_scanned += 1
+                result = self._analyze_file(fp, service_id, lang)
+                for m in result["methods"]:
+                    all_methods[m["id"]] = m
+                all_calls.extend(result["calls"])
+
+        # Persist method nodes (batched)
+        method_nodes = []
         for method_id, info in all_methods.items():
             node = MethodNode(
                 id=method_id,
@@ -351,12 +370,14 @@ class MethodGraphBuilder:
                 service_id=service_id,
                 typed_signature=info.get("typed_signature"),
             )
-            self.store.upsert_node(node)
+            method_nodes.append(node)
+        self.store.upsert_nodes_batch(method_nodes)
 
-        # Build and persist call graph edges
+        # Build and persist call graph edges (batched)
         call_graph = self._build_call_graph(all_methods, all_calls)
+        edge_objects = []
         for caller_id, callee_id, meta in call_graph:
-            self.store.upsert_edge(
+            edge_objects.append(
                 DependencyEdge(
                     source_id=caller_id,
                     target_id=callee_id,
@@ -364,6 +385,7 @@ class MethodGraphBuilder:
                     metadata=meta,
                 )
             )
+        self.store.upsert_edges_batch(edge_objects)
 
         return {
             "methods": len(all_methods),

corbell/core/graph/sqlite_store.py

@@ -134,6 +134,31 @@ class SQLiteGraphStore(GraphStore):
             )
             conn.commit()
 
+    def upsert_nodes_batch(self, nodes: list) -> None:
+        """Batch-upsert multiple nodes in a single transaction."""
+        if not nodes:
+            return
+        with self._conn() as conn:
+            conn.executemany(
+                "INSERT OR REPLACE INTO graph_nodes (id, node_type, data) VALUES (?, ?, ?)",
+                [(node.id, _node_type_str(node), json.dumps(_node_to_dict(node))) for node in nodes],
+            )
+            conn.commit()
+
+    def upsert_edges_batch(self, edges: list) -> None:
+        """Batch-upsert multiple edges in a single transaction."""
+        if not edges:
+            return
+        with self._conn() as conn:
+            conn.executemany(
+                """INSERT INTO graph_edges (source_id, target_id, kind, metadata)
+                   VALUES (?, ?, ?, ?)
+                   ON CONFLICT(source_id, target_id, kind)
+                   DO UPDATE SET metadata = excluded.metadata""",
+                [(e.source_id, e.target_id, e.kind, json.dumps(e.metadata)) for e in edges],
+            )
+            conn.commit()
+
     def _load_node(self, row) -> ServiceNode | DataStoreNode | QueueNode | MethodNode:
         return _dict_to_node(row["node_type"], json.loads(row["data"]))