feat: include property definitions per label in get_graph_schema

Adds per-label/type property key discovery via json_each() so users can
see which properties are available for Cypher queries without trial and
error. Base columns listed first, JSON keys appended (capped at 50).

Closes #179.
Based on #181 by dLo999.

Co-Authored-By: Dustin Obrecht <dustin@kurtnoble.com>

Author: test

README.md

@@ -317,7 +317,7 @@ codebase-memory-mcp cli --raw search_graph '{"label": "Function"}' | jq '.result
 | `trace_call_path` | BFS traversal — who calls a function and what it calls. Depth 1-5. |
 | `detect_changes` | Map git diff to affected symbols + blast radius with risk classification. |
 | `query_graph` | Execute Cypher-like graph queries (read-only). |
-| `get_graph_schema` | Node/edge counts, relationship patterns. Run this first. |
+| `get_graph_schema` | Node/edge counts, relationship patterns, property definitions per label. Run this first. |
 | `get_code_snippet` | Read source code for a function by qualified name. |
 | `get_architecture` | Codebase overview: languages, packages, routes, hotspots, clusters, ADR. |
 | `search_code` | Grep-like text search within indexed project files. |

src/mcp/mcp.c

@@ -994,6 +994,11 @@ static char *handle_get_graph_schema(cbm_mcp_server_t *srv, const char *args) {
         yyjson_mut_val *lbl = yyjson_mut_obj(doc);
         yyjson_mut_obj_add_str(doc, lbl, "label", schema.node_labels[i].label);
         yyjson_mut_obj_add_int(doc, lbl, "count", schema.node_labels[i].count);
+        yyjson_mut_val *props = yyjson_mut_arr(doc);
+        for (int j = 0; j < schema.node_labels[i].property_count; j++) {
+            yyjson_mut_arr_add_str(doc, props, schema.node_labels[i].properties[j]);
+        }
+        yyjson_mut_obj_add_val(doc, lbl, "properties", props);
         yyjson_mut_arr_add_val(labels, lbl);
     }
     yyjson_mut_obj_add_val(doc, root, "node_labels", labels);
@@ -1003,6 +1008,11 @@ static char *handle_get_graph_schema(cbm_mcp_server_t *srv, const char *args) {
         yyjson_mut_val *typ = yyjson_mut_obj(doc);
         yyjson_mut_obj_add_str(doc, typ, "type", schema.edge_types[i].type);
         yyjson_mut_obj_add_int(doc, typ, "count", schema.edge_types[i].count);
+        yyjson_mut_val *eprops = yyjson_mut_arr(doc);
+        for (int j = 0; j < schema.edge_types[i].property_count; j++) {
+            yyjson_mut_arr_add_str(doc, eprops, schema.edge_types[i].properties[j]);
+        }
+        yyjson_mut_obj_add_val(doc, typ, "properties", eprops);
         yyjson_mut_arr_add_val(types, typ);
     }
     yyjson_mut_obj_add_val(doc, root, "edge_types", types);

src/store/store.c

@@ -2696,6 +2696,36 @@ int cbm_deduplicate_hops(const cbm_node_hop_t *hops, int hop_count, cbm_node_hop
 
 /* ── Schema ─────────────────────────────────────────────────────── */
 
+enum { SCHEMA_MAX_JSON_KEYS = 50 };
+
+/* Discover distinct JSON property keys for a table/column via json_each().
+ * Prepends base_cols, then appends up to SCHEMA_MAX_JSON_KEYS from the query.
+ * Caller must free the returned array and each string in it. */
+static void schema_discover_props(sqlite3 *db, const char *sql, const char *project,
+                                  const char *filter, const char **base_cols, int base_col_count,
+                                  char ***out_props, int *out_count) {
+    int pcap = base_col_count + SCHEMA_MAX_JSON_KEYS;
+    char **props = malloc(pcap * sizeof(char *));
+    int pn = 0;
+
+    for (int b = 0; b < base_col_count; b++) {
+        props[pn++] = heap_strdup(base_cols[b]);
+    }
+
+    sqlite3_stmt *pstmt = NULL;
+    if (sqlite3_prepare_v2(db, sql, CBM_NOT_FOUND, &pstmt, NULL) == SQLITE_OK) {
+        bind_text(pstmt, SKIP_ONE, project);
+        bind_text(pstmt, PAIR_LEN, filter);
+        while (sqlite3_step(pstmt) == SQLITE_ROW && pn < pcap) {
+            props[pn++] = heap_strdup((const char *)sqlite3_column_text(pstmt, 0));
+        }
+        sqlite3_finalize(pstmt);
+    }
+
+    *out_props = props;
+    *out_count = pn;
+}
+
 int cbm_store_get_schema(cbm_store_t *s, const char *project, cbm_schema_info_t *out) {
     memset(out, 0, sizeof(*out));
     if (!s || !s->db) {
@@ -2720,13 +2750,34 @@ int cbm_store_get_schema(cbm_store_t *s, const char *project, cbm_schema_info_t
             }
             arr[n].label = heap_strdup((const char *)sqlite3_column_text(stmt, 0));
             arr[n].count = sqlite3_column_int(stmt, SKIP_ONE);
+            arr[n].properties = NULL;
+            arr[n].property_count = 0;
             n++;
         }
         sqlite3_finalize(stmt);
         out->node_labels = arr;
         out->node_label_count = n;
     }
 
+    /* Node label property keys: base columns + distinct JSON property keys per label */
+    {
+        static const char *node_base_cols[] = {"name", "qualified_name", "file_path", "start_line",
+                                               "end_line"};
+        const char *prop_sql = "SELECT DISTINCT je.key "
+                               "FROM nodes, json_each(nodes.properties) AS je "
+                               "WHERE nodes.project = ?1 AND nodes.label = ?2 "
+                               "  AND nodes.properties != '{}' "
+                               "ORDER BY je.key "
+                               "LIMIT 50;";
+
+        for (int i = 0; i < out->node_label_count; i++) {
+            schema_discover_props(
+                s->db, prop_sql, project, out->node_labels[i].label, node_base_cols,
+                (int)(sizeof(node_base_cols) / sizeof(node_base_cols[0])),
+                &out->node_labels[i].properties, &out->node_labels[i].property_count);
+        }
+    }
+
     /* Edge types */
     {
         const char *sql = "SELECT type, COUNT(*) FROM edges WHERE project = ?1 GROUP BY type ORDER "
@@ -2745,13 +2796,33 @@ int cbm_store_get_schema(cbm_store_t *s, const char *project, cbm_schema_info_t
             }
             arr[n].type = heap_strdup((const char *)sqlite3_column_text(stmt, 0));
             arr[n].count = sqlite3_column_int(stmt, SKIP_ONE);
+            arr[n].properties = NULL;
+            arr[n].property_count = 0;
             n++;
         }
         sqlite3_finalize(stmt);
         out->edge_types = arr;
         out->edge_type_count = n;
     }
 
+    /* Edge type property keys: base columns + distinct JSON property keys per type */
+    {
+        static const char *edge_base_cols[] = {"source_id", "target_id"};
+        const char *prop_sql = "SELECT DISTINCT je.key "
+                               "FROM edges, json_each(edges.properties) AS je "
+                               "WHERE edges.project = ?1 AND edges.type = ?2 "
+                               "  AND edges.properties != '{}' "
+                               "ORDER BY je.key "
+                               "LIMIT 50;";
+
+        for (int i = 0; i < out->edge_type_count; i++) {
+            schema_discover_props(s->db, prop_sql, project, out->edge_types[i].type, edge_base_cols,
+                                  (int)(sizeof(edge_base_cols) / sizeof(edge_base_cols[0])),
+                                  &out->edge_types[i].properties,
+                                  &out->edge_types[i].property_count);
+        }
+    }
+
     return CBM_STORE_OK;
 }
 
@@ -2761,11 +2832,19 @@ void cbm_store_schema_free(cbm_schema_info_t *out) {
     }
     for (int i = 0; i < out->node_label_count; i++) {
         free((void *)out->node_labels[i].label);
+        for (int j = 0; j < out->node_labels[i].property_count; j++) {
+            free(out->node_labels[i].properties[j]);
+        }
+        free(out->node_labels[i].properties);
     }
     free(out->node_labels);
 
     for (int i = 0; i < out->edge_type_count; i++) {
         free((void *)out->edge_types[i].type);
+        for (int j = 0; j < out->edge_types[i].property_count; j++) {
+            free(out->edge_types[i].properties[j]);
+        }
+        free(out->edge_types[i].properties);
     }
     free(out->edge_types);
 

src/store/store.h

@@ -163,11 +163,15 @@ typedef struct {
 typedef struct {
     const char *label;
     int count;
+    char **properties; /* distinct property keys for this label (base + JSON) */
+    int property_count;
 } cbm_label_count_t;
 
 typedef struct {
     const char *type;
     int count;
+    char **properties; /* distinct property keys for this edge type (base + JSON) */
+    int property_count;
 } cbm_type_count_t;
 
 typedef struct {