pi-harness-runtime 1.1.413 → 1.1.415

package/scripts/sync-to-qdrant.ts

@@ -28,60 +28,16 @@ import {
 import {
   LEGACY_SKILL_COLLECTION,
   SKILL_COLLECTION,
+  skillPointPayload,
+  skillQdrantPoint,
 } from "../packages/skills/dist/vector-schema.js";
-
-// =============================================================================
-// SANITIZATION: Replace client-specific names with placeholders
-// =============================================================================
-
-interface SanitizationRule {
-  pattern: RegExp;
-  replacement: string;
-}
-
-const SANITIZATION_RULES: SanitizationRule[] = [
-  // Client company names (case-insensitive)
-  { pattern: /\binpac\b/gi, replacement: "[Client A]" },
-  { pattern: /\bm-capital\b/gi, replacement: "[Client A]" },
-  { pattern: /\bmcapital\b/gi, replacement: "[Client A]" },
-  { pattern: /\bdigisoft\b/gi, replacement: "[Client B]" },
-
-  // Project/app specific names
-  { pattern: /\bgse-insurance\b/gi, replacement: "[Insurance Project]" },
-  { pattern: /\btbs-import\b/gi, replacement: "[Import Project]" },
-  { pattern: /\bdigisoft_erp\b/gi, replacement: "[Client B Erp]" },
-  { pattern: /\bm_capital\b/gi, replacement: "[Client A]" },
-  { pattern: /\binpac_\b/gi, replacement: "[Client A]_" },
-  { pattern: /\bpaperclip_\b/gi, replacement: "[Client D]_" },
-  { pattern: /\bopenclaw_\b/gi, replacement: "[Client F]_" },
-
-  // Generic replacements
-  { pattern: /Company Name/gi, replacement: "[Client Name]" },
-  { pattern: /your company/gi, replacement: "[Client]" },
-  { pattern: /our client/gi, replacement: "[Client]" },
-  { pattern: /client-specific/gi, replacement: "[Project-specific]" },
-];
-
-// Skill name sanitization (for skill directory name -> sanitized display name)
-const SKILL_NAME_MAP: Record<string, string> = {
-  'gse-insurance-step5-slot-filter': 'insurance-project-step-filter',
-  'inpac-pe-approval-workflow': 'client-a-pe-approval-workflow',
-  'inpac-po-approval-workflow': 'client-a-po-approval-workflow',
-  'inpac-pr-approval-workflow': 'client-a-pr-approval-workflow',
-  'tbs-import-clearance-lcv': 'import-project-clearance',
-  'whispertool': 'whisper-tool',
-};
-
-/**
- * Sanitize text by replacing client-specific patterns
- */
-function sanitize(text: string): string {
-  let result = text;
-  for (const rule of SANITIZATION_RULES) {
-    result = result.replace(rule.pattern, rule.replacement);
-  }
-  return result;
-}
+// The sanitization lives in the shared package: the runtime registry writes
+// into the same collection and needs the same rules, or it would leave the
+// raw name and the raw body of a client skill in a shared cluster.
+import {
+  sanitizeText,
+  skillPointName,
+} from "../packages/skills/dist/sanitize.js";
 
 // =============================================================================
 // Qdrant Config
@@ -123,7 +79,6 @@ function parseHostPort(url: string): { host: string; port: number } {
 // =============================================================================
 
 interface SkillDoc {
-  id: number;
   name: string;
   description: string;
   body: string;
@@ -156,23 +111,24 @@ function readSkill(dirPath: string): SkillDoc | null {
     const originalName = fm.name || basename(dirPath);
     
     // Sanitize content
-    const sanitizedContent = sanitize(rawContent);
+    const sanitizedContent = sanitizeText(rawContent);
     
     // Use name map for known sensitive skill names, otherwise sanitize
-    const name = SKILL_NAME_MAP[originalName] || sanitize(originalName);
+    const name = skillPointName(originalName);
     
-    const description = sanitize(fm.description || "");
+    const description = sanitizeText(fm.description || "");
     const body = sanitizedContent.replace(/^---[\s\S]*?---\n/, "").trim();
-    
-    // Text for embedding (sanitized)
-    const text = `${name}. ${description}. ${body.slice(0, 1000)}`;
+
+    // The payload and the text come from the shared schema, so this script
+    // and the runtime store the SAME point for the same skill: the point
+    // id is the hash of the name the payload carries.
+    const payload = skillPointPayload({ name, description, body });
 
     return {
-      id: 0,
-      name,
-      description,
-      body: body.slice(0, 2000),
-      text,
+      name: payload.name,
+      description: payload.description,
+      body: payload.body,
+      text: payload.text,
     };
   } catch {
     return null;
@@ -273,12 +229,16 @@ async function upsertSkills(
   }
 
   const points = skills.map((skill, idx) => ({
-    id: idx + 1,
-    payload: {
+    // The point comes from the shared schema, so this script and the
+    // runtime store the SAME point for the same skill. A sequential id
+    // would leave a second point for every skill the runtime also indexes,
+    // and could overwrite an unrelated one.
+    ...skillQdrantPoint({
       name: skill.name,
       description: skill.description,
       body: skill.body,
-    },
+      text: skill.text,
+    }),
     vector: embeddings[idx],
   }));
 
@@ -431,7 +391,7 @@ Usage:
 Environment:
   QDRANT_API_KEY    Your Qdrant Cloud API key
   QDRANT_CLUSTER_ENDPOINT  Qdrant URL
-  QDRANT_COLLECTION Collection name (default: pi-harness-skills)
+  QDRANT_COLLECTION Collection name (default: ${SKILL_COLLECTION})
 
 Or create file:
   ~/.pi-harness-runtime/keys/qdrant-api-key.txt