@@ -2,6 +2,16 @@
 
 ## Unreleased
 

@@ -0,0 +204,4 @@
+| `status` | `String(20)` | NOT NULL, DEFAULT "pending", CHECK IN (`pending`, `indexing`, `ready`, `stale`, `error`) |
+| `error_message` | `Text` | NULLABLE |
+| `created_at` | `String(30)` | NOT NULL (ISO-8601 UTC) |
+

@@ -0,0 +1,917 @@
+"""Repository indexing service for CleverAgents.

@@ -0,0 +268,4 @@
+
+        # Check whether a previous good index exists.  If so, we do NOT
+        # overwrite it with an INDEXING placeholder — that would destroy the
+        # previous index if the walk fails (NEW-1 fix).  For fresh resources

@@ -0,0 +500,4 @@
+        return repo_index
+
+    def get_index(self, resource_id: str) -> RepoIndex | None:
+        """Retrieve the persisted index for a resource.

@@ -0,0 +675,4 @@
+
+                # Skip paths that exceed FileRecord.path max_length (1024)
+                # to avoid a Pydantic ValidationError that would crash the
+                # entire index (R4-DEF1 fix).

@@ -3095,0 +3099,4 @@
+# ---------------------------------------------------------------------------
+
+
+class RepoIndexModel(Base):

@@ -3095,0 +3127,4 @@
+    error_message = Column(Text, nullable=True)
+
+    # Timestamps (ISO-8601 strings, UTC)
+    indexed_at = Column(String(30), nullable=False)

@@ -0,0 +71,4 @@
+
+    def time_incremental_refresh_no_changes(self) -> None:
+        """Incremental refresh when no files changed (100 files)."""
+        self.svc.index_resource(_RID, self.tmpdir)

@@ -0,0 +189,4 @@
+| `context.ignore_patterns` | `ContextConfig.ignore_patterns` | Exclude globs |
+| `context.max_file_size` | `ContextConfig.max_file_size` | Per-file size limit |
+| `context.max_total_size` | `ContextConfig.max_total_size` | Total index size cap |
+| `context.indexing_strategy` | `ContextConfig.indexing_strategy` | `full` or `incremental` |

@@ -0,0 +228,4 @@
+- Lines 2829-2916: `project link-resource` triggers indexing
+- Lines 3322-3395: `project show` displays index status
+- Lines 19719-19727: Project data model index fields
+- Lines 28649-28664: `index.*` configuration keys

@@ -0,0 +508,4 @@
+    """Assert a file's hash changed after modification."""
+    original: RepoIndex = context.repo_index_original_result  # type: ignore[attr-defined]
+    current: RepoIndex = context.repo_index_result  # type: ignore[attr-defined]
+    orig_hash = next(f.content_hash for f in original.files if f.path == filename)

@@ -0,0 +83,4 @@
+    try:
+        # Initial index
+        original = svc.index_resource(_RID, tmpdir)
+        orig_hash = next(f.content_hash for f in original.files if f.path == "app.py")

@@ -0,0 +12,4 @@
+Full Index With Language Detection And Persistence
+    [Documentation]    Index a sample directory, verify file count, language, and persistence.
+    [Tags]    feature195
+    ${result}=    Run Process    ${PYTHON}    ${HELPER}    full-index    cwd=${WORKSPACE}

@@ -141,0 +148,4 @@
+    from sqlalchemy import create_engine
+    from sqlalchemy.orm import sessionmaker
+
+    engine = create_engine(database_url, echo=False)

@@ -0,0 +128,4 @@
+        session.add(db_index)
+        session.flush()
+
+        # Bulk insert file records

@@ -0,0 +174,4 @@
+                token_count=cast(int, fr.token_count),
+                language=cast(str, fr.language),
+                size_bytes=cast(int, fr.size_bytes),
+                last_modified=datetime.fromisoformat(cast(str, fr.last_modified)),

@@ -0,0 +187,4 @@
+            token_estimate=cast(int, row.token_estimate),
+            primary_language=cast(str, row.primary_language),
+            status=IndexStatus(cast(str, row.status)),
+            error_message=(

@@ -0,0 +146,4 @@
+        if not root.is_dir():
+            raise ValueError(f"Resource root path is not a directory: {root}")
+
+        index_id = str(ULID())

@@ -0,0 +152,4 @@
+        # overwrite it with an INDEXING placeholder — that would destroy the
+        # previous index if the walk fails (NEW-1 fix).
+        existing_index = self.get_index(resource_id)
+        if existing_index is None:

@@ -0,0 +186,4 @@
+                        index_id=index_id,
+                        resource_id=resource_id,
+                        status=IndexStatus.ERROR,
+                        error_message=str(exc),

@@ -0,0 +358,4 @@
+        )
+
+        repo_index = RepoIndex(metadata=metadata, files=tuple(merged))
+        persist_index(self._session_factory, repo_index)

@@ -0,0 +173,4 @@
+    """
+    # Exclude takes precedence
+    for pattern in exclude_globs:
+        if fnmatch.fnmatch(rel_path, pattern):

@@ -0,0 +230,4 @@
+    # Phase 1: Collect all candidate files (path, stat) sorted by rel_path
+    candidates: list[tuple[str, Path, os.stat_result]] = []
+
+    for dirpath, dirnames, filenames in os.walk(root):

@@ -0,0 +276,4 @@
+            except OSError:
+                continue
+
+            if (

@@ -0,0 +298,4 @@
+        if (
+            max_total_size is not None
+            and max_total_size > 0
+            and total_bytes + size_bytes > max_total_size

@@ -0,0 +309,4 @@
+        # Hash content — then re-stat to avoid TOCTOU: the file may have
+        # changed between the Phase-1 stat() and this read.
+        try:
+            content_hash = hash_file(full_path)

@@ -0,0 +82,4 @@
+    Based on specification.md project data model (lines 19719-19727).
+    """
+
+    path: str = Field(

@@ -0,0 +88,4 @@
+        max_length=1024,
+        description="Relative path within the resource root",
+    )
+    content_hash: str = Field(

@@ -0,0 +98,4 @@
+        ge=0,
+        description="Estimated token count for the file",
+    )
+    language: str = Field(

@@ -0,0 +112,4 @@
+        description="Last modification timestamp (UTC)",
+    )
+
+    @field_validator("last_modified", mode="before")

@@ -3095,0 +3137,4 @@
+            "status IN ('pending', 'indexing', 'ready', 'stale', 'error')",
+            name="ck_repo_indexes_status",
+        ),
+        Index("ix_repo_indexes_resource_id", "resource_id"),