feat(investigation): publish incremental reference graphs

Publish graph deltas using the Server-owned cumulative graph contract. Validate historical comparison chart publications and consolidate Agent documentation.

.opencode/tools/publish_task_progress.ts

@@ -1,16 +1,16 @@
 import { tool } from "../tool.js";
 import { getInternalBridgeConfig } from "../internalBridge.js";
 import { taskProgressSchema } from "../../src/product/investigation-report.js";
-import { taskPublicationProgressSchema } from "./task-publication-schema.js";
+import { taskPublicationGraphProgressSchema } from "./task-publication-schema.js";
 
 export default tool({
-  description: "Publish the complete current investigation progress from a task execution session. Publish before starting a real investigation step and when changing direction. The summary and optional Mermaid diagram describe actual activity for people; they never schedule the Agent or derive status from todo. Every node_refs node_id must exactly reuse an id declared in the Mermaid source; do not use A/B nodes in Mermaid and separate semantic ids in node_refs. The entire diagram may be redrawn, and only explicitly running nodes indicate current activity. Map targets reference known supply features. Use a stable UUID request_id and reuse it on retries. Publishing progress does not complete the run or confirm any human work.",
+  description: "Save an optional investigation process reference with summary and optional graph_update. Update when the route materially changes, an important finding or failed attempt should be recorded, or a report is being organized. No per-step publication, frequent refresh, or live synchronization is required. The Server retains prior nodes and paths and generates Mermaid. Use stable node_id values; only submit changes. Before extending a saved graph, read get_task_context section=progress when needed. Optional status annotations are reference, recorded, needs_input, failed, and not_taken. They describe the saved explanation, not execution lifecycle: recorded means a finding exists, not that a task, calculation, work order, or field action completed. Retain failed attempts with the actual reason in detail; append a new node and edge for a retry or alternative. Text-only updates preserve the graph. Graphs never schedule execution, settle runs, or infer human completion. Use known supply feature ids and saved evidence/action references. Reuse request_id on retries.",
   args: {
     request_id: tool.schema.string().uuid().describe("Stable UUID idempotency key; reuse for retries of this progress publication"),
-    progress: taskPublicationProgressSchema,
+    progress: taskPublicationGraphProgressSchema,
   },
   async execute(args, context) {
-    const progress = taskProgressSchema.parse(args.progress);
+    const progress = taskProgressSchema.parse(taskPublicationGraphProgressSchema.parse(args.progress));
     const { baseUrl, internalToken } = getInternalBridgeConfig();
     const response = await fetch(`${baseUrl}/internal/operational-task/artifacts`, {
       method: "POST",