fix(update): run incremental graph rebuild, drop stale "not implemented" warning (#310)

`java-codebase-rag update` did a Lance-only catch-up then unconditionally
printed "AST graph (LadybugDB) incremental rebuild is not yet implemented",
even though incremental graph shipped (67a76de) and `increment` already calls
run_incremental_graph. The two commands share the word "update" but use
different code paths; only the `increment` path (cli.py) was rewired — the
installer's run_update index step was never updated.

Now run_update mirrors `increment`: after run_cocoindex_update it calls
run_incremental_graph, so both index layers stay current. The stale warning
is removed entirely; on graph-step failure it prints a truthful, actionable
message instead of claiming the feature is missing. The warning constant
stays in cli.py for the legitimate `increment --vectors-only` opt-out path.

Docs/help (cli --help, README, JAVA-CODEBASE-RAG-CLI.md) updated to describe
update as artifact refresh + incremental Lance/graph catch-up, and to drop the
false "prints graph staleness warning" line.

Co-authored-by: Claude <noreply@anthropic.com>

Author: HumanBean17

README.md

@@ -63,7 +63,7 @@ java-codebase-rag install
 java-codebase-rag install --non-interactive --agent claude-code
 ```
 
-After `pip install --upgrade java-codebase-rag`, run `java-codebase-rag update` to refresh shipped artifacts.
+After `pip install --upgrade java-codebase-rag`, run `java-codebase-rag update` to refresh shipped artifacts and catch up the index (Lance + graph).
 
 ### Manual registration
 
@@ -195,7 +195,7 @@ Run `java-codebase-rag --help` to list grouped subcommands. Operator playbook wi
 | Group | Subcommand | What it does |
 |---|---|---|
 | Setup | `install` | Interactive setup wizard: config, MCP registration, skill/agent deployment, indexing. |
-| Setup | `update` | Refresh shipped artifacts (skill, agent, MCP entry) after pip upgrade. |
+| Setup | `update` | Refresh shipped artifacts (skill, agent, MCP entry) + incremental Lance/graph catch-up after pip upgrade. |
 | Lifecycle | `init` | First-time index. Refuses if artifacts already exist. |
 | Lifecycle | `increment` | CocoIndex catch-up + incremental Kuzu update. `--vectors-only` for Lance only. |
 | Lifecycle | `reprocess` | Full Lance + Kuzu rebuild. `--vectors-only` / `--graph-only` for a single phase. |

docs/JAVA-CODEBASE-RAG-CLI.md

@@ -61,7 +61,7 @@ java-codebase-rag install --scope user
 
 ### `update`
 
-Post-upgrade refresh: overwrites skill and agent files with the latest shipped versions and updates the MCP command path. Requires a prior `install` run.
+Post-upgrade refresh: overwrites skill and agent files with the latest shipped versions and updates the MCP command path. If an index exists, also runs an incremental Lance + graph catch-up (same as `increment`). Requires a prior `install` run.
 
 ```bash
 # Refresh after pip upgrade
@@ -83,7 +83,7 @@ java-codebase-rag update --force
 - Detects previously configured agent hosts (scans both project-level and user-level config files).
 - Refreshes skill and agent files (versioned assets from the package).
 - Updates MCP entrypoint path if `java-codebase-rag-mcp` has moved.
-- Runs `increment` on the index if it exists (LanceDB catch-up). Prints graph staleness warning.
+- Runs an incremental index update (Lance + graph) if an index exists — same as `java-codebase-rag increment`.
 - Skips MCP config if the entry already exists and is correct.
 
 **Exit codes:**

java_codebase_rag/cli.py

@@ -775,8 +775,9 @@ def build_parser() -> argparse.ArgumentParser:
         help="Refresh shipped artifacts (skill, agent, MCP entry) after pip upgrade.",
         description=(
             "Post-upgrade refresh: overwrites skill and agent files with the latest "
-            "shipped versions and updates the MCP command path. Use --dry-run to "
-            "preview changes without writing. Requires a prior `install` run."
+            "shipped versions and updates the MCP command path. If an index exists, "
+            "also runs an incremental Lance + graph catch-up (same as `increment`). "
+            "Use --dry-run to preview changes without writing. Requires a prior `install` run."
         ),
     )
     update.add_argument(

java_codebase_rag/installer.py

@@ -1182,7 +1182,7 @@ def run_update(
         index_dir_has_existing_artifacts,
         resolve_operator_config,
     )
-    from java_codebase_rag.pipeline import run_cocoindex_update
+    from java_codebase_rag.pipeline import run_cocoindex_update, run_incremental_graph
 
     project_root = discover_project_root(cwd)
     if project_root is None:
@@ -1207,22 +1207,37 @@ def run_update(
         print("Run `java-codebase-rag install` to create one.")
         return EXIT_PARTIAL if has_artifact_failures else EXIT_SUCCESS
 
-    # Run increment (LanceDB catch-up)
+    # Run increment: LanceDB catch-up + incremental graph rebuild.
+    # Mirrors `java-codebase-rag increment` so both index layers stay current.
+    # The "graph not implemented" warning belongs only on the vectors-only path
+    # (increment --vectors-only), where the graph step is deliberately skipped.
     if not dry_run:
-        print("\nUpdating index (incremental LanceDB update)...")
+        print("\nUpdating index (Lance + graph)...")
         cfg.apply_to_os_environ()
         env = cfg.subprocess_env()
 
         coco = run_cocoindex_update(env, full_reprocess=False, quiet=True)
         if coco.returncode != 0:
-            print(f"Error: Index update failed with code {coco.returncode}")
+            print(f"Error: Lance index update failed with code {coco.returncode}")
             return 1
 
-        # Print graph staleness warning
-        from java_codebase_rag.cli import _INCREMENT_WARNING_LINES
-        print("\n" + "\n".join(_INCREMENT_WARNING_LINES))
+        g = run_incremental_graph(
+            source_root=cfg.source_root,
+            ladybug_path=cfg.ladybug_path,
+            verbose=False,
+            quiet=True,
+            env=env,
+        )
+        if g.returncode != 0:
+            # Artifacts above already refreshed; the graph catch-up is best-effort
+            # here. Surface a truthful, actionable message instead of leaving the
+            # graph silently stale or claiming the feature is unimplemented.
+            print(
+                f"\nWarning: incremental graph update failed (exit {g.returncode}). "
+                "Run `java-codebase-rag reprocess` for a full rebuild."
+            )
     else:
-        print("\nWould run incremental index update.")
+        print("\nWould run incremental index update (Lance + graph).")
 
     # Print summary
     print("\nUpdate complete.")

tests/test_installer.py

@@ -1106,6 +1106,10 @@ def mock_run_build_ast_graph(*args, **kwargs):
             from subprocess import CompletedProcess
             return CompletedProcess(["build_ast_graph"], 0)
 
+        def mock_run_incremental_graph(*args, **kwargs):
+            from subprocess import CompletedProcess
+            return CompletedProcess(["build_ast_graph", "--incremental"], 0)
+
         monkeypatch.setattr(
             "java_codebase_rag.pipeline.run_cocoindex_update",
             mock_run_cocoindex_update,
@@ -1114,6 +1118,10 @@ def mock_run_build_ast_graph(*args, **kwargs):
             "java_codebase_rag.pipeline.run_build_ast_graph",
             mock_run_build_ast_graph,
         )
+        monkeypatch.setattr(
+            "java_codebase_rag.pipeline.run_incremental_graph",
+            mock_run_incremental_graph,
+        )
 
         # Change to fixture directory
         monkeypatch.setattr(Path, "cwd", lambda: cwd)