Update OpenClaw OrgX prompts for MCP v2

docs/launch/demo-script-saturday.md

@@ -37,7 +37,7 @@ Goal: show the full flow from zero to visible autonomous progress, end-to-end.
 5. Ask "what should I do next?"
    - In chat: "What should I do next?"
    - Expect:
-     - `recommend_next_action` includes active agent/stream progress (e.g. "Agents running: marketing (25%) ...")
+     - `orgx_recommend` includes active agent/stream progress (e.g. "Agents running: marketing (25%) ...")
      - Suggested parallel workstream to review.
 
 6. Completion + artifact visibility

docs/marketing/manual-agent-dispatch-golden-prompt.md

@@ -61,8 +61,8 @@ Workflow (do not skip):
   - docs/marketing/live/* (if relevant)
   - docs/orgx-openclaw-launch-workstreams-plan-2026-02-14.md (if relevant)
 - If OrgX MCP tools are available:
-  - mcp__orgx__query_org_memory (prior messaging, constraints, decisions)
-  - mcp__orgx__list_entities (existing campaign/content artifacts to avoid duplication)
+  - mcp__orgx__orgx_search (prior messaging, constraints, decisions)
+  - mcp__orgx__orgx_search (existing campaign/content artifacts to avoid duplication)
 4) Draft the artifact as JSON first (no prose-first).
 5) Validate and fix until clean:
 python3 /Users/hopeatina/.codex/skills/marketing-agent/scripts/validate_marketing.py <artifact_file> --type <campaign|content|sequence>

docs/ops/2026-02-25/org-memory-context-survival-discovery.md

@@ -20,7 +20,7 @@ Discovery-only slice to verify whether Org Memory returns real data today and wh
 
 ### G3.1 (Org Memory Returns Real Data)
 
-1. There is no dedicated MCP tool for querying org memory/history (for example `query_org_memory`) in current tool registration; memory visibility is indirect through `orgx_sync` and live snapshot endpoints.
+1. There is no dedicated MCP tool for querying org memory/history (for example `orgx_search`) in current tool registration; memory visibility is indirect through `orgx_sync` and live snapshot endpoints.
 2. `orgx_sync` currently acts as a thin pass-through and does not enrich or expose a normalized memory-focused view for operators.
 3. The background sync loop only sends local `agents` telemetry in `syncMemory` calls, so text memory/daily logs are not auto-refreshed unless explicitly pushed by a caller.
 

qa-artifacts/milestone-breakdown-verify/api-data.json

@@ -1909,7 +1909,7 @@
           "tasks": [
             {
               "id": "d9a7293a-cd64-4d43-ba99-7919bbfbd07c",
-              "title": "Test entity_action lifecycle (launch → pause → complete) on each entity type",
+              "title": "Test orgx_act lifecycle (launch → pause → complete) on each entity type",
               "status": "todo"
             },
             {
@@ -1929,7 +1929,7 @@
             },
             {
               "id": "177f9786-edb7-4abe-8ab6-88b6b8e36391",
-              "title": "Test score_next_up_queue returns ranked items with factor breakdowns",
+              "title": "Test orgx_recommend returns ranked items with factor breakdowns",
               "status": "todo"
             },
             {

src/contracts/client.ts

@@ -593,7 +593,7 @@ export class OrgXClient {
     limit?: number;
   }): Promise<Record<string, unknown>> {
     return this.executeClientTool<Record<string, unknown>>(
-      "query_org_memory",
+      "orgx_search",
       params
     );
   }
@@ -607,7 +607,7 @@ export class OrgXClient {
     cascade?: boolean;
   }): Promise<Record<string, unknown>> {
     return this.executeClientTool<Record<string, unknown>>(
-      "recommend_next_action",
+      "orgx_recommend",
       params
     );
   }