@nanmicoder/dsh-agent-teams 0.1.13 → 0.1.14

package/lib/types/types.d.ts

@@ -11,8 +11,51 @@
 export type TaskStatus = 'pending' | 'claimed' | 'in_progress' | 'completed' | 'failed' | 'cancelled';
 /** Statuses after which a task can no longer be claimed or worked on. */
 export declare const TERMINAL_TASK_STATUSES: readonly TaskStatus[];
+/** Structured quality-gate kind. Absent / unknown values are treated as `work`. */
+export type TaskKind = 'requirements' | 'implementation' | 'verification' | 'review' | 'repair' | 'integration' | 'work';
+export declare const TASK_KINDS: readonly TaskKind[];
+/** Review / requirements conclusion. Only `pass` may complete those kinds. */
+export type ReviewVerdict = 'pass' | 'needs_revision' | 'reject';
+export declare const REVIEW_VERDICTS: readonly ReviewVerdict[];
+/** Finding severity used by review / requirements output. */
+export type FindingSeverity = 'low' | 'medium' | 'high' | 'blocker';
+export declare const FINDING_SEVERITIES: readonly FindingSeverity[];
+/** One structured review finding. */
+export interface ReviewFinding {
+    /** Stable id, for example `SEC-001`. */
+    id: string;
+    severity: FindingSeverity;
+    file?: string;
+    line?: number;
+    problem: string;
+    requiredFix: string;
+    resolved?: boolean;
+}
+/** One acceptance criterion result recorded at completion. */
+export interface AcceptanceResult {
+    criterion: string;
+    status: 'passed' | 'failed';
+    evidence?: string;
+}
+/** One verification command result recorded at completion. */
+export interface CommandResult {
+    command: string;
+    status: 'passed' | 'failed';
+    exitCode?: number;
+    evidence?: string;
+}
+/** Profile / team review-loop limits. */
+export interface ReviewPolicy {
+    requirementsMinRounds?: number;
+    requirementsMaxRounds?: number;
+    codeMaxRounds?: number;
+    maxRepairAttempts?: number;
+    requiredReviewers?: string[];
+}
 /** One task of a team's task list. */
 export interface TeamTask {
+    /** Stable task id from the profile template; absent for ad-hoc tasks. */
+    profileSeedId?: string;
     /** Stable task id within the team (`t1`, `t2`, …). */
     id: string;
     /** Brief title for the task. */
@@ -34,6 +77,29 @@ export interface TeamTask {
     handoffId?: string;
     /** A handoff is quiescing the old owner; the scheduler must not dispatch it yet. */
     reassigning?: boolean;
+    /** Quality-gate kind. Missing values are treated as `work`. */
+    kind?: TaskKind;
+    /** Review / requirements / repair loop index, 1-based when present. */
+    round?: number;
+    verdict?: ReviewVerdict;
+    findings?: ReviewFinding[];
+    objective?: string;
+    inScope?: string[];
+    outOfScope?: string[];
+    acceptance?: string[];
+    verify?: string[];
+    deliverables?: string[];
+    nonGoals?: string[];
+    changedPaths?: string[];
+    acceptanceResults?: AcceptanceResult[];
+    commandsRun?: CommandResult[];
+    reviewedTaskId?: string;
+    reviewedAttempt?: number;
+    /** Repair source: the implementation / previous successful artifact. */
+    sourceTaskId?: string;
+    sourceFindingIds?: string[];
+    /** User-constraint / goal items this task claims to cover. */
+    coverageOf?: string[];
     createdAt: number;
     updatedAt: number;
 }
@@ -53,6 +119,15 @@ export interface TeamMember {
     model?: string;
     /** Resolved reasoning effort captured from the captain or target model default. */
     reasoningEffort?: string;
+    /** Prompt specific to this member's execution turns. */
+    executionPrompt?: string;
+    /** Configured second-choice route. */
+    fallback?: TeamModelFallback;
+    /** Active route after fallback, without changing the primary descriptor route. */
+    activeProvider?: string;
+    activeModel?: string;
+    /** Whether the fallback route is currently active. */
+    fallbackActive?: boolean;
     joinedAt: number;
     status: MemberStatus;
 }
@@ -72,6 +147,22 @@ export interface TeamMessage {
     /** Set once the recipient has consumed or been shown the durable fallback. */
     readAt?: number;
 }
+/** Snapshot of the named profile used to seed a team. */
+export interface TeamModelFallback {
+    provider: string;
+    model: string;
+}
+export interface TeamProfileSnapshot {
+    name: string;
+    description?: string;
+    protocol?: string;
+    executionPrompt?: string;
+    fallback?: TeamModelFallback;
+    /** Frozen planning mode: captain plans the graph; seed keeps template tasks. */
+    taskPlanning?: 'captain' | 'seed';
+    /** Frozen review-loop policy from the creating profile. */
+    reviewPolicy?: ReviewPolicy;
+}
 /** The full durable team record. */
 export interface TeamState {
     /** Original team name. */
@@ -80,6 +171,8 @@ export interface TeamState {
     id: string;
     /** Team purpose/goal. */
     description?: string;
+    /** Immutable named profile snapshot, when created from a profile. */
+    profile?: TeamProfileSnapshot;
     /** Session id of the captain agent that owns this team. */
     captainSessionId: string;
     createdAt: number;
@@ -88,4 +181,29 @@ export interface TeamState {
     tasks: TeamTask[];
     /** Monotonic task id counter. */
     taskSeq: number;
+    /**
+     * Two-phase execution lifecycle. Missing means `running` for durable
+     * compatibility with teams created before staging existed.
+     */
+    phase?: 'staged' | 'running';
+    /**
+     * Human-facing review sub-state while `phase` is `staged`. Missing staged
+     * records are treated as `awaiting_review` for backward compatibility.
+     * `awaiting_feedback` means the user returned to chat and the Captain must
+     * ask what should change before editing this same draft.
+     */
+    planReviewState?: 'awaiting_review' | 'awaiting_feedback';
+    /** Timestamp written only after a staged plan is explicitly approved. */
+    approvedAt?: number;
+    /**
+     * Human halt from the captain chat. The team remains on disk, members stay
+     * available, and unfinished work is cancelled until the captain resumes.
+     */
+    halted?: boolean;
+    /** Timestamp of the latest human halt, when present. */
+    haltedAt?: number;
+    /** Review-loop policy snapshot copied from the creating profile, when present. */
+    reviewPolicy?: ReviewPolicy;
+    /** Set when an automatic review/repair loop hits its configured ceiling. */
+    escalated?: boolean;
 }

package/lib/types.js

@@ -9,3 +9,14 @@
  */
 /** Statuses after which a task can no longer be claimed or worked on. */
 export const TERMINAL_TASK_STATUSES = ['completed', 'failed', 'cancelled'];
+export const TASK_KINDS = [
+    'requirements',
+    'implementation',
+    'verification',
+    'review',
+    'repair',
+    'integration',
+    'work',
+];
+export const REVIEW_VERDICTS = ['pass', 'needs_revision', 'reject'];
+export const FINDING_SEVERITIES = ['low', 'medium', 'high', 'blocker'];

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nanmicoder/dsh-agent-teams",
-  "version": "0.1.13",
+  "version": "0.1.14",
   "description": "AgentTeams for DeepSeek Harness: multi-agent team collaboration (captain, members, tasks with dependencies, messaging) driven by natural language, with a tree monitor in the web GUI",
   "type": "module",
   "main": "lib/index.js",
@@ -67,17 +67,18 @@
         "@deepseek-ai/dsh-client-locale",
         "@deepseek-ai/dsh-client-runtime",
         "@deepseek-ai/dsh-client-ui-conversation",
-        "@deepseek-ai/dsh-client-ui-layout"
+        "@deepseek-ai/dsh-client-ui-layout",
+        "@deepseek-ai/dsh-client-ui-model-selection"
       ],
       "platform": "web"
     }
   },
   "scripts": {
-    "build": "tsc -p tsconfig.json && tsc -p tsconfig.client.json && tsdown",
+    "build": "node scripts/clean-build.mjs && tsc -p tsconfig.json && tsc -p tsconfig.client.json && tsdown",
     "typecheck": "tsc -p tsconfig.json --noEmit && tsc -p tsconfig.client.json --noEmit",
     "sync:skill": "node scripts/sync-skill.mjs",
     "verify:skill": "node scripts/sync-skill.mjs --check",
-    "verify": "node scripts/verify.mjs && node scripts/lifecycle-verify.mjs && node scripts/stress-verify.mjs && pnpm verify:skill",
+    "verify": "node scripts/verify.mjs && node scripts/fallback-tdd.mjs && node scripts/quality-gates-tdd.mjs && node scripts/lifecycle-verify.mjs && node scripts/stress-verify.mjs && pnpm verify:skill",
     "prepublishOnly": "pnpm build && pnpm verify"
   },
   "peerDependencies": {
@@ -87,6 +88,7 @@
     "@deepseek-ai/dsh-client-runtime": "^0.1.0-rc.6",
     "@deepseek-ai/dsh-client-ui-conversation": "^0.1.0-rc.6",
     "@deepseek-ai/dsh-client-ui-layout": "^0.1.0-rc.6",
+    "@deepseek-ai/dsh-client-ui-model-selection": "^0.1.0-rc.8",
     "@deepseek-ai/dsh-client-ui-primitives": "^0.1.0-rc.6",
     "@deepseek-ai/dsh-client-ui-slots": "^0.1.0-rc.6",
     "@deepseek-ai/dsh-commands": "^0.1.0-rc.6",
@@ -117,6 +119,9 @@
     "@deepseek-ai/dsh-client-ui-layout": {
       "optional": true
     },
+    "@deepseek-ai/dsh-client-ui-model-selection": {
+      "optional": true
+    },
     "@deepseek-ai/dsh-client-ui-primitives": {
       "optional": true
     },
@@ -155,6 +160,7 @@
     "@deepseek-ai/dsh-client-runtime": "0.1.0-rc.8",
     "@deepseek-ai/dsh-client-ui-conversation": "0.1.0-rc.8",
     "@deepseek-ai/dsh-client-ui-layout": "0.1.0-rc.8",
+    "@deepseek-ai/dsh-client-ui-model-selection": "0.1.0-rc.8",
     "@deepseek-ai/dsh-client-ui-primitives": "0.1.0-rc.8",
     "@deepseek-ai/dsh-client-ui-slots": "0.1.0-rc.8",
     "@deepseek-ai/dsh-commands": "0.1.0-rc.8",

package/release-notes/v0.1.14.md

@@ -0,0 +1,68 @@
+# AgentTeams v0.1.14
+
+This release adds a review-first team workflow, stronger delivery gates, and safer controls for long-running AgentTeams sessions.
+
+## New
+
+- **Review plans before execution**: normal `/agent-teams` runs now stage the roster and task DAG without creating members or scheduling work. Reviewers can edit members, prompts, models, reasoning levels, tasks, and dependencies before launch.
+- **Native model routing**: the plan editor uses Harness's provider/model catalog and reasoning metadata instead of maintaining a separate handwritten model list.
+- **Captain-designed teams and stop controls**: profiles can let the Captain design the task graph for the real workspace, while running teams expose a dedicated, confirmed stop action in the activity panel.
+- **Quality delivery gates**: optional requirements, implementation, verification, independent review, repair, re-review, and integration contracts keep downstream work blocked until evidence is accepted.
+
+## Fixed & Improved
+
+- **Atomic approval**: every member route is validated before any child session is created, preventing a bad model choice from leaving a partially started team.
+- **Return to chat and revise**: returning from review cancels the planning turn, asks one focused question, and applies the answer to the same draft with one atomic plan edit before requesting review again.
+- **Discard really stops planning**: discarding archives the draft, aborts an active Captain turn, and injects authoritative context that prevents the model from silently recreating the team.
+- **Responsive review actions**: decision controls respond to the activity panel's own width, so approve, revise, and discard remain aligned even when the panel is narrow.
+- **Safer recovery and takeover**: abandoned Captain-owned tasks return to the scheduler without allowing stale attempts to overwrite newer work.
+- **More stable panel state**: restored activity panels preserve their collapsed state and retain complete staged, running, stopped, and archived histories.
+
+## Verification
+
+- Passed TypeScript checks, production builds, offline verification, lifecycle verification, quality-gate TDD, the complex stress suite, and package dry-run checks.
+- Exercised the staged-plan workflow in the real Harness Web UI with DeepSeek-V4-Flash / High: native model routing, narrow-panel layout, return-to-chat revision, atomic plan editing, discard cancellation, archive state, and ordinary chat after discard all passed.
+- Verified that discarding while the Captain was still planning stopped the active generation immediately and did not recreate the team.
+
+## Installation
+
+```sh
+dsh plugin --profile web add @nanmicoder/dsh-agent-teams
+```
+
+<details>
+<summary><b>中文版本 / Chinese Version</b></summary>
+
+# AgentTeams v0.1.14
+
+本次更新加入执行前审查流程、更强的交付质量门禁，以及更安全的长任务控制。
+
+## 新功能
+
+- **执行前审查计划**：普通 `/agent-teams` 会先生成可编辑的成员阵容和任务 DAG，不创建成员、不调度任务。启动前可调整成员、提示词、模型、推理等级、任务和依赖。
+- **复用 Harness 模型路由**：计划编辑器直接使用 Harness 的 Provider、模型目录和推理等级元数据，不再维护另一份手写模型列表。
+- **队长规划与团队停止**：profile 可让队长根据真实 workspace 设计任务图；运行中的团队可从活动面板执行带确认的停止操作。
+- **交付质量门禁**：可选的需求、实现、验证、独立审查、修复、复审和集成合同，会在证据验收前持续阻断下游任务。
+
+## 修复与改进
+
+- **原子启动校验**：创建任何子会话之前先校验全部成员模型路由，避免错误模型造成“半支团队”。
+- **返回对话修改**：返回聊天会取消当前规划轮次，只追问一个修改问题；用户回复后通过一次原子计划编辑更新同一份草案，再次等待审查。
+- **放弃后真正停止**：放弃计划会归档草案、中止仍在运行的队长轮次，并注入禁止静默重建团队的权威上下文。
+- **窄面板操作稳定**：操作区根据活动面板自身宽度响应，面板再窄也不会让确认、修改和放弃按钮错位重叠。
+- **更安全的恢复与接管**：被遗留的队长任务会安全回到调度器，旧 attempt 不能覆盖较新的执行结果。
+- **面板状态更稳定**：恢复后的活动面板保持折叠状态，并完整保留待审、运行、停止和归档历史。
+
+## 验证
+
+- 通过 TypeScript 检查、生产构建、离线验证、生命周期验证、质量门禁 TDD、复杂压力测试和 npm 包 dry-run。
+- 在真实 Harness Web UI 中使用 DeepSeek-V4-Flash / High 完成执行前计划回归：原生模型路由、窄面板、返回对话、原子修改、放弃取消、归档状态及放弃后的普通聊天均通过。
+- 验证队长仍在规划时放弃：生成立即停止，团队不会被重新创建。
+
+## 安装
+
+```sh
+dsh plugin --profile web add @nanmicoder/dsh-agent-teams
+```
+
+</details>