@xopcai/xopc 0.0.410 → 0.0.411

package/dist/src/workflows/store/run-store.js

@@ -1,40 +0,0 @@
-import { projectWorkflowRunView } from "../engine/projector.js";
-import { WorkflowEventStore } from "./event-store.js";
-import { WorkflowRunIndexStore } from "./run-index-store.js";
-//#region src/workflows/store/run-store.ts
-var WorkflowRunStore = class {
-	eventStore;
-	indexStore = new WorkflowRunIndexStore();
-	constructor(config, agentId, eventStore) {
-		this.agentId = agentId;
-		this.eventStore = eventStore ?? new WorkflowEventStore(config, agentId);
-	}
-	async rebuildRunView(runId) {
-		const view = projectWorkflowRunView(await this.eventStore.readRunEvents(runId));
-		if (!view) return null;
-		this.indexStore.upsert(this.agentId, view);
-		return view;
-	}
-	async readRunView(runId) {
-		return this.rebuildRunView(runId);
-	}
-	async listRunSummaries(limit = 50, options = {}) {
-		const safeLimit = Math.min(500, Math.max(1, Math.floor(limit)));
-		return this.indexStore.list(this.agentId, {
-			limit: safeLimit,
-			projectId: options.projectId
-		});
-	}
-	async listRunSummariesForTask(taskId, limit = 50) {
-		const safeLimit = Math.min(500, Math.max(1, Math.floor(limit)));
-		return this.indexStore.list(this.agentId, {
-			taskId,
-			limit: safeLimit
-		});
-	}
-};
-function createWorkflowRunStore(config, agentId) {
-	return new WorkflowRunStore(config, agentId);
-}
-//#endregion
-export { WorkflowRunStore, createWorkflowRunStore };

package/skills/engineering/workflow-authoring/SKILL.md

@@ -1,32 +0,0 @@
----
-name: workflow-authoring
-description: Create or revise an executable xopc workflow directly in a normal conversation using a durable validated draft.
-metadata:
-  i18n:
-    en:
-      name: "Workflow Authoring"
-      description: "Design or revise an executable XOPC workflow through a durable conversational draft."
-    zh-CN:
-      name: "工作流创作"
-      description: "通过持久化对话草稿设计或修改可执行的 XOPC 工作流。"
-  xopc:
-    emoji: "◆"
-    requires_tools:
-      - workflow_manage
----
-
-# Workflow authoring
-
-Use this skill whenever the user asks to create, change, duplicate, or improve a workflow in an ordinary conversation.
-
-1. If the request may refer to an existing workflow, call `workflow_manage` with `list` before choosing a name.
-2. Call `workflow_manage` with `open_draft` and the chosen workflow name. Keep the returned draft id and `updatedAtMs` for later calls.
-3. Translate the user's outcome into the smallest connected acyclic graph that can deliver it. Prefer one capable agent step over several vague steps.
-4. Keep exactly one input and one output. Use decisions only for real branching and merge nodes only for real fan-in.
-5. Write agent prompts as complete operating instructions. State inputs, expected result, constraints, and how predecessor results are used.
-6. Add `manifest.resources.skills` only when a runtime step genuinely needs that skill. Use `skills_list` and `skill_view` to verify its exact name and instructions.
-7. Apply the complete graph and manifest with `workflow_manage` action `replace_draft`, always passing the latest `updatedAtMs` as `expectedUpdatedAtMs`. If validation rejects it, fix the reported issue instead of explaining it away.
-8. Validate with `workflow_manage` action `validate_draft`, then summarize the steps, assumptions, required access, and important tradeoffs in plain language.
-9. Publish with `workflow_manage` action `publish_draft` only after explicit user confirmation, passing `confirmed: true` and the latest draft version as `expectedUpdatedAtMs`.
-
-Never emit raw workflow JSON for the user to copy, never claim an unsaved proposal is durable, and never publish merely because the draft validates.