codevira 1.7.0__tar.gz → 1.7.1__tar.gz

{codevira-1.7.0 → codevira-1.7.1}/CHANGELOG.md

@@ -13,6 +13,37 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 
 ---
 
+## [1.7.1] — 2026-04-22 — Search Timeout Fix & Version Display
+
+Two small but user-visible fixes on top of v1.7.0.
+
+### Fixed
+
+- **`search_codebase` timeout on first call** (reported by a user testing
+  on Antigravity). The embedding model (`all-MiniLM-L6-v2`) was being
+  instantiated fresh on every MCP tool call, which triggered a ~90MB
+  download + PyTorch init on first ever use (30-60s on slow networks)
+  and 1-3s of re-init overhead on every subsequent call. Antigravity's
+  ~30s MCP tool timeout killed the query before the model finished loading.
+
+  Three-layer fix:
+  1. Module-level cache for the chroma client + embedding function,
+     keyed by `db_dir`. Subsequent calls are now instant.
+  2. Background `prewarm_embedding_model()` spawned at server startup
+     (both stdio and HTTP transports). Model loads in parallel with
+     the MCP handshake window.
+  3. Cold-path timeout guard: if a query arrives while warmup is still
+     in progress, returns `{"status": "warming", ...}` within 10 seconds
+     instead of blocking until the MCP timeout fires. The agent gets a
+     clean retryable response.
+
+- **`codevira register` banner showed hardcoded `v1.6`** after upgrading
+  to v1.7.0. Now reads `mcp_server.__version__` dynamically. Same fix
+  applied to `metadata.json` version field written during auto-init and
+  migration.
+
+---
+
 ## [1.7.0] — 2026-04-18 — Token Efficiency & AI-First Tool Design
 
 **The biggest release since v1.0.** We realized Codevira was dumping 15k-60k

{codevira-1.7.0/codevira.egg-info → codevira-1.7.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: codevira
-Version: 1.7.0
+Version: 1.7.1
 Summary: Persistent adaptive memory for AI coding agents — MCP server with context graph, semantic search, adaptive learning, roadmap tracking, and cross-tool continuity.
 Author-email: Sachin Shelke <sachin.worldnet@gmail.com>
 License: MIT
@@ -44,7 +44,7 @@ Provides-Extra: all
 [![Python](https://img.shields.io/badge/python-3.10%2B-blue)](https://www.python.org/)
 [![License: MIT](https://img.shields.io/badge/license-MIT-green)](LICENSE)
 [![MCP](https://img.shields.io/badge/protocol-MCP-purple)](https://modelcontextprotocol.io)
-[![Version](https://img.shields.io/badge/version-1.7.0-orange)](CHANGELOG.md)
+[![Version](https://img.shields.io/badge/version-1.7.1-orange)](CHANGELOG.md)
 [![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen)](CONTRIBUTING.md)
 
 **Built for solo developers** working on local projects with AI agents. Codevira gives every AI tool you use access to the same persistent project memory — so you stop re-explaining your codebase every session, stop losing carefully-made decisions, and stop burning tokens on re-discovery.

{codevira-1.7.0 → codevira-1.7.1}/README.md

@@ -5,7 +5,7 @@
 [![Python](https://img.shields.io/badge/python-3.10%2B-blue)](https://www.python.org/)
 [![License: MIT](https://img.shields.io/badge/license-MIT-green)](LICENSE)
 [![MCP](https://img.shields.io/badge/protocol-MCP-purple)](https://modelcontextprotocol.io)
-[![Version](https://img.shields.io/badge/version-1.7.0-orange)](CHANGELOG.md)
+[![Version](https://img.shields.io/badge/version-1.7.1-orange)](CHANGELOG.md)
 [![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen)](CONTRIBUTING.md)
 
 **Built for solo developers** working on local projects with AI agents. Codevira gives every AI tool you use access to the same persistent project memory — so you stop re-explaining your codebase every session, stop losing carefully-made decisions, and stop burning tokens on re-discovery.

{codevira-1.7.0 → codevira-1.7.1/codevira.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: codevira
-Version: 1.7.0
+Version: 1.7.1
 Summary: Persistent adaptive memory for AI coding agents — MCP server with context graph, semantic search, adaptive learning, roadmap tracking, and cross-tool continuity.
 Author-email: Sachin Shelke <sachin.worldnet@gmail.com>
 License: MIT
@@ -44,7 +44,7 @@ Provides-Extra: all
 [![Python](https://img.shields.io/badge/python-3.10%2B-blue)](https://www.python.org/)
 [![License: MIT](https://img.shields.io/badge/license-MIT-green)](LICENSE)
 [![MCP](https://img.shields.io/badge/protocol-MCP-purple)](https://modelcontextprotocol.io)
-[![Version](https://img.shields.io/badge/version-1.7.0-orange)](CHANGELOG.md)
+[![Version](https://img.shields.io/badge/version-1.7.1-orange)](CHANGELOG.md)
 [![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen)](CONTRIBUTING.md)
 
 **Built for solo developers** working on local projects with AI agents. Codevira gives every AI tool you use access to the same persistent project memory — so you stop re-explaining your codebase every session, stop losing carefully-made decisions, and stop burning tokens on re-discovery.

{codevira-1.7.0 → codevira-1.7.1}/mcp_server/__init__.py

@@ -14,4 +14,4 @@ internals — internal APIs may change between minor versions.
 from mcp_server.cli import main
 
 __all__ = ["main"]
-__version__ = "1.7.0"
+__version__ = "1.7.1"

{codevira-1.7.0 → codevira-1.7.1}/mcp_server/auto_init.py

@@ -223,6 +223,14 @@ def _write_config(data_dir: Path, detected: dict, project_root: Path) -> None:
         yaml.dump(config, f, default_flow_style=False, sort_keys=False)
 
 
+def _codevira_version() -> str:
+    try:
+        from mcp_server import __version__
+        return __version__
+    except Exception:
+        return "unknown"
+
+
 def _write_metadata(data_dir: Path, project_root: Path) -> None:
     """Write metadata.json so the centralized dir is recognized."""
     import json
@@ -234,7 +242,7 @@ def _write_metadata(data_dir: Path, project_root: Path) -> None:
         "git_remote": _get_git_remote_url(project_root),
         "original_path": str(project_root),
         "created_at": datetime.now(timezone.utc).isoformat(),
-        "version": "1.6.0",
+        "version": _codevira_version(),
         "auto_initialized": True,
     }
     (data_dir / "metadata.json").write_text(json.dumps(metadata, indent=2))

{codevira-1.7.0 → codevira-1.7.1}/mcp_server/cli.py

@@ -464,8 +464,9 @@ def cmd_register(
     project_root = get_project_root()
     cmd_path, python_exe = _resolve_command()
 
+    from mcp_server import __version__
     print()
-    print("  Codevira — Global IDE Registration (v1.6)")
+    print(f"  Codevira — Global IDE Registration (v{__version__})")
     print("  " + "─" * 44)
     print()
 

{codevira-1.7.0 → codevira-1.7.1}/mcp_server/http_server.py

@@ -340,6 +340,13 @@ def run_http_server(
     except Exception as e:
         logger.warning("Log retention cleanup failed: %s", e)
 
+    # v1.7: Pre-warm the embedding model in a background thread
+    try:
+        from mcp_server.tools.search import prewarm_embedding_model
+        prewarm_embedding_model()
+    except Exception as e:
+        logger.warning("Embedding prewarm failed: %s", e)
+
     # ---- TLS certificate setup ----
     ssl_certfile: str | None = None
     ssl_keyfile: str | None = None

{codevira-1.7.0 → codevira-1.7.1}/mcp_server/migrate.py

@@ -22,7 +22,12 @@ from pathlib import Path
 
 logger = logging.getLogger(__name__)
 
-_CODEVIRA_VERSION = "1.6.0"
+def _codevira_version() -> str:
+    try:
+        from mcp_server import __version__
+        return __version__
+    except Exception:
+        return "unknown"
 
 
 def detect_migration_needed(project_root: Path) -> bool:
@@ -140,7 +145,7 @@ def migrate_to_centralized(project_root: Path) -> dict:
         "git_remote": git_remote,
         "original_path": str(project_root),
         "migrated_at": datetime.now(timezone.utc).isoformat(),
-        "version": _CODEVIRA_VERSION,
+        "version": _codevira_version(),
     }
     (centralized / "metadata.json").write_text(json.dumps(metadata, indent=2))
 

{codevira-1.7.0 → codevira-1.7.1}/mcp_server/server.py

@@ -1135,6 +1135,15 @@ def main():
         try: log_crash(e, context="log retention cleanup")
         except Exception: pass
 
+    # v1.7: Pre-warm the embedding model in a background thread so the first
+    # search_codebase() call doesn't hit the MCP client's ~30s timeout
+    # while waiting for PyTorch init / model download.
+    try:
+        from mcp_server.tools.search import prewarm_embedding_model
+        prewarm_embedding_model()
+    except Exception as e:
+        logger.warning("Embedding prewarm failed: %s", e)
+
     async def _run():
         async with mcp.server.stdio.stdio_server() as (read_stream, write_stream):
             await server.run(

{codevira-1.7.0 → codevira-1.7.1}/mcp_server/tools/search.py

@@ -9,6 +9,16 @@ def _get_db() -> SQLiteGraph:
     db_path = get_data_dir() / "graph" / "graph.db"
     return SQLiteGraph(db_path)
 
+
+# Module-level cache for the semantic search stack.
+# Creating SentenceTransformerEmbeddingFunction triggers:
+#   - ~90MB model download on first use (cached in ~/.cache/huggingface/)
+#   - PyTorch/onnxruntime init (~1-3s even after download)
+# Caching avoids paying this cost on every search_codebase call — first
+# call is still slow but subsequent calls are instant.
+_chroma_cache: dict = {"client": None, "embed_fn": None, "db_dir": None}
+
+
 def _get_chroma_client():
     try:
         import chromadb
@@ -20,12 +30,48 @@ def _get_chroma_client():
     if not db_dir.exists():
         return None, None
 
+    # Reuse cached client + embed_fn if we're hitting the same project's index
+    # (PersistentClient is tied to a specific path; invalidate if path changes).
+    if (
+        _chroma_cache["client"] is not None
+        and _chroma_cache["embed_fn"] is not None
+        and _chroma_cache["db_dir"] == str(db_dir)
+    ):
+        return _chroma_cache["client"], _chroma_cache["embed_fn"]
+
     client = chromadb.PersistentClient(path=str(db_dir))
     embed_fn = embedding_functions.SentenceTransformerEmbeddingFunction(
         model_name="all-MiniLM-L6-v2"
     )
+    _chroma_cache["client"] = client
+    _chroma_cache["embed_fn"] = embed_fn
+    _chroma_cache["db_dir"] = str(db_dir)
     return client, embed_fn
 
+
+def prewarm_embedding_model() -> None:
+    """Pre-load the embedding model in a background thread at server startup.
+
+    SentenceTransformerEmbeddingFunction triggers a ~90MB model download on
+    first use and 1-3s of PyTorch init. If we wait until the first
+    search_codebase() call, Antigravity's ~30s MCP timeout can kill the query.
+
+    Called from server.main() and http_server.run_http_server() via a daemon
+    thread — the server becomes ready immediately; search gets warmed up in
+    parallel. Safe to call multiple times (cached after first successful load).
+    """
+    import threading
+
+    def _warmup():
+        try:
+            _get_chroma_client()  # populates _chroma_cache
+        except Exception:
+            # Failure is non-fatal — search_codebase will retry on next call
+            pass
+
+    t = threading.Thread(target=_warmup, daemon=True, name="codevira-embed-prewarm")
+    t.start()
+
 def search_codebase(description: str, top_k: int = 5, include_content: bool = False) -> dict[str, Any]:
     """Semantic search over the codebase.
 
@@ -36,6 +82,39 @@ def search_codebase(description: str, top_k: int = 5, include_content: bool = Fa
     Pass include_content=True only when you explicitly need the chunk source
     inline (can be 500-3000 tokens per match).
     """
+    # Fast path: if another thread is still pre-warming the embedding model,
+    # return a "warming" status immediately instead of blocking for 30s+ and
+    # triggering the MCP client's timeout. Agent retries get served once
+    # warmup is done.
+    if _chroma_cache["embed_fn"] is None:
+        # Cache miss. Try to populate it (fast if model is already downloaded
+        # and cached on disk; slow only the first time ever across any project).
+        # Wrap in a short timeout so we don't block the MCP thread.
+        import threading as _th
+        load_done = _th.Event()
+        load_err = [None]
+
+        def _load():
+            try:
+                _get_chroma_client()
+            except Exception as e:
+                load_err[0] = e
+            finally:
+                load_done.set()
+
+        _th.Thread(target=_load, daemon=True, name="codevira-embed-sync-load").start()
+        # Wait up to 10 seconds — enough for warm-cache loads, short of MCP timeout
+        if not load_done.wait(timeout=10.0):
+            return {
+                "status": "warming",
+                "message": (
+                    "Semantic search model is loading (first-time setup downloads ~90MB). "
+                    "Try this query again in 30-60 seconds. Other tools work normally."
+                ),
+            }
+        if load_err[0] is not None:
+            return {"error": f"Embedding model load failed: {load_err[0]}"}
+
     client, embed_fn = _get_chroma_client()
     if not client:
         # v1.6: Check if auto-init is running and return a friendly status

{codevira-1.7.0 → codevira-1.7.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "codevira"
-version = "1.7.0"
+version = "1.7.1"
 description = "Persistent adaptive memory for AI coding agents — MCP server with context graph, semantic search, adaptive learning, roadmap tracking, and cross-tool continuity."
 readme = "README.md"
 license = { text = "MIT" }

{codevira-1.7.0 → codevira-1.7.1}/tests/test_auto_init.py

@@ -330,7 +330,8 @@ class TestWriteMetadata:
         assert meta["path_key"] == "test_key"
         assert meta["git_remote"] == "git@github.com:test/repo.git"
         assert meta["original_path"] == str(project_root)
-        assert meta["version"] == "1.6.0"
+        from mcp_server import __version__
+        assert meta["version"] == __version__
         assert meta["auto_initialized"] is True
         assert "created_at" in meta
 

{codevira-1.7.0 → codevira-1.7.1}/tests/test_migrate.py

@@ -237,7 +237,8 @@ class TestMigrateToCentralized:
         meta = json.loads(meta_file.read_text())
         assert meta["path_key"] == key
         assert meta["original_path"] == str(project)
-        assert meta["version"] == "1.6.0"
+        from mcp_server import __version__
+        assert meta["version"] == __version__
         assert "migrated_at" in meta
 
     def test_metadata_contains_git_remote(self, tmp_path, monkeypatch):