feat: Set review_code include/exclude paths in config file (#130)

* feat: ability to set review_code include/exclude paths in config file

* feat(docs): review_code_paths

* fix(docs): trim trailing whitespaces

Author: sttlr

docs/config/review_code_paths.md

@@ -0,0 +1,63 @@
+# Selective Code Review: Include/Exclude Paths
+
+Metis now supports fine-grained control over which files are targeted during a code review. By utilizing `review_code_include_paths` and `review_code_exclude_paths` in your configuration, you can focus the engine on critical business logic while skipping boilerplate, definitions, and tests.
+
+## Overview
+
+These settings allow you to filter files **specifically for the review process** without removing them from the project's broader context. This ensures that Metis stays fast and the results remain relevant.
+
+- **`review_code_include_paths`**: Limits the review to specific directories or files.
+- **`review_code_exclude_paths`**: Explicitly skips specific files or patterns during the review phase.
+
+The **`review_code_include_paths`** and **`review_code_exclude_paths`** configurations utilize standard **gitignore-style** pattern matching.
+
+---
+
+## Why use this instead of `.metisignore`?
+
+It is important to distinguish between these configuration options and the global `.metisignore` file:
+
+| Feature | Scope | Impact on Context |
+| :--- | :--- | :--- |
+| **`.metisignore`** | **Global** | Files are completely invisible. They are not indexed and cannot be used by the model at all (e.g., `.venv`, `dist`, `node_modules`). |
+| **Review Paths** | **Command-Specific** | Files are skipped by the `review_code` logic, but **remain available** as "Relevant Context" (via embeddings or tools) to help the AI understand the code it *is* reviewing. |
+
+**The Logic:** You generally don't want Metis to waste time looking for bugs in a `.d.ts` or `interface.ts` file. However, if Metis is reviewing a service that imports those types, it still needs to be able to "see" those files to understand the data structures.
+
+---
+
+## Configuration
+
+Add these options to your `metis.yaml` file under the `metis_engine` block.
+
+### Example Configuration
+
+```yaml
+metis_engine:
+  # Only run reviews within the backend folder
+  # Where the Core Logic resides
+  review_code_include_paths:
+    - 'backend/'
+
+  # Skip files that don't require logic analysis
+  review_code_exclude_paths:
+    - 'dto/'
+    - '*.d.ts'
+    - '*.spec.ts'
+    - '*.fixture.ts'
+    - '*.dto.ts'
+    - '*.interface.ts'
+    - '*.type.ts'
+    - '*.schema.ts'
+    - 'backend/test/'
+    - 'e2e/'
+```
+
+View other examples for focused review at the `examples/focused_review_code` folder.
+
+## Key Benefits
+
+- **Focused Reviews**: Configure Metis to only review Core Logic, suchs as specific apps inside your big monorepo.
+- **Significant Speed Increase**: Metis avoids reviewing definition files and tests.
+- **Noise Reduction**: Prevents "false positive" issues or irrelevant suggestions on generated code, type interfaces, or DTOs where logic-based review is unnecessary.
+- **Preserved Context**: Unlike a global ignore, the AI still has access to these files to provide better context for the logic it is actually reviewing.

examples/focused_review_code/metis_typescript_example.yaml

@@ -0,0 +1,29 @@
+metis_engine:
+
+  # Only review code inside the "backend" folder
+  review_code_include_paths:
+    - 'backend/'
+
+  # Don't review dtos, schemas, tests...
+  review_code_exclude_paths:
+    - 'dto/'
+    - '*.d.ts'
+    - '*.spec.ts'
+    - '*.fixture.ts'
+    - '*.dto.ts'
+    - '*.const.ts'
+    - '*.interface.ts'
+    - '*.model.ts'
+    - '*.type.ts'
+    - '*.request.ts'
+    - '*.response.ts'
+    - '*.response.api.ts'
+    - '*.params.ts'
+    - '*.schema.ts'
+    - '*.mapper.ts'
+    - '*.store.ts'
+    - 'e2e_ssr/'
+    - 'e2e/'
+    - '*.e2e-spec.ts'
+    - 'backend/test'
+    - 'schema.ts'

src/metis/configuration.py

@@ -127,6 +127,12 @@ def load_runtime_config(config_path=None, enable_psql=False):
         },
     )
     runtime["metisignore_file"] = engine_cfg.get("metisignore_file", None)
+    runtime["review_code_include_paths"] = engine_cfg.get(
+        "review_code_include_paths", []
+    )
+    runtime["review_code_exclude_paths"] = engine_cfg.get(
+        "review_code_exclude_paths", []
+    )
 
     # Query config
     query_cfg = cfg.get("query", {})

src/metis/engine/core.py

@@ -93,23 +93,25 @@ def __init__(
         self._review_graph = None
         self._ask_graph = None
         self.metisignore_file = kwargs.get("metisignore_file") or ".metisignore"
+        self.review_code_include_paths = kwargs.get("review_code_include_paths", [])
+        self.review_code_exclude_paths = kwargs.get("review_code_exclude_paths", [])
 
-    def load_metisignore(self):
+    def load_metisignore(self) -> pathspec.GitIgnoreSpec | None:
         """
-        Load metisignore file and return a PathSpec matcher.
+        Load metisignore file and return a GitIgnoreSpec matcher.
 
         Args:
             metisignore: Path to a file that have the ignore regex ( use the .gitignore syntax )
 
         Returns:
-            pathspec.PathSpec object or None if file doesn't exist
+            pathspec.GitIgnoreSpec object or None if file doesn't exist
         """
         try:
             if not self.metisignore_file:
                 logger.info("No MetisIgnore file provided")
                 return None
             with open(self.metisignore_file, "r") as f:
-                spec = pathspec.PathSpec.from_lines("gitwildmatch", f)
+                spec = pathspec.GitIgnoreSpec.from_lines(f)
                 logger.info(f"MetisIgnore file loaded: {self.metisignore_file}")
             return spec
         except FileNotFoundError:
@@ -340,19 +342,35 @@ def review_file(self, file_path):
     def get_code_files(self):
         """
         Return a list of file names in the self.codebase_path folder.
-        Evaulate the path with metisignore file if requested
+        Evaluate the path with metisignore file, include/exclude paths if requested
         """
         base_path = os.path.abspath(self.codebase_path)
         metisignore_spec = self.load_metisignore()
+        include_spec = None
+        if self.review_code_include_paths:
+            include_spec = pathspec.GitIgnoreSpec.from_lines(
+                self.review_code_include_paths
+            )
+        exclude_spec = None
+        if self.review_code_exclude_paths:
+            exclude_spec = pathspec.GitIgnoreSpec.from_lines(
+                self.review_code_exclude_paths
+            )
         file_list = []
         for root, _, files in os.walk(base_path):
             for file in files:
+                full_path = os.path.join(root, file)
                 ext = os.path.splitext(file)[1].lower()
-                if ext in self.code_exts and (
-                    not metisignore_spec
-                    or not metisignore_spec.match_file(os.path.join(root, file))
-                ):
-                    file_list.append(os.path.join(root, file))
+                if ext not in self.code_exts:
+                    continue
+                rel_path = os.path.relpath(full_path, base_path)
+                if metisignore_spec and metisignore_spec.match_file(rel_path):
+                    continue
+                if include_spec and not include_spec.match_file(rel_path):
+                    continue
+                if exclude_spec and exclude_spec.match_file(rel_path):
+                    continue
+                file_list.append(full_path)
         return file_list
 
     def review_code(self):