@mstar-harness/omp 3.6.0-alpha.4 → 3.6.1

package/agents/architect.md

@@ -1,8 +1,6 @@
 ---
 name: architect
-description: |-
-  技术架构师 - 系统设计、技术决策与技术向文档编写（架构说明、ADR、接口契约等）。
-  Architect - system design, technical decisions, and technical documentation (architecture notes, ADRs, interface contracts).
+description: "Architect - system design, technical decisions, and technical documentation (architecture notes, ADRs, interface contracts)."
 mode: subagent
 tools:
   write: true

package/agents/code-reviewer.md

@@ -1,8 +1,6 @@
 ---
 name: code-reviewer
-description: |-
-  代码审查员 - SDD 任务级审查（L2）+ 代码库审计执行（audit）。只读席位：不实现、不修代码、不占 QC 席。
-  Code Reviewer - SDD per-task review (L2) + codebase audit execution (audit). Read-only seat: does not implement, fix, or occupy a QC seat.
+description: "Code Reviewer - SDD per-task review (L2) + codebase audit execution (audit). Read-only seat: does not implement, fix, or occupy a QC seat."
 mode: subagent
 tools:
   write: true

package/agents/frontend-dev.md

@@ -1,8 +1,6 @@
 ---
 name: frontend-dev
-description: |-
-  前端开发工程师 - 页面/组件/交互/a11y/前端性能。全栈功能里默认前端主责（与 `@fullstack-dev` 分轨）；纯 UI 任务首选本角色。
-  Frontend Developer - pages/components/interactions/accessibility/frontend performance. This is the default frontend owner in fullstack work (split with `@fullstack-dev`) and the preferred role for pure UI tasks.
+description: "Frontend Developer - pages/components/interactions/accessibility/frontend performance. This is the default frontend owner in fullstack work (split with `@fullstack-dev`) and the preferred role for pure UI tasks."
 mode: subagent
 tools:
   write: true

package/agents/fullstack-dev-2.md

@@ -1,8 +1,6 @@
 ---
 name: fullstack-dev-2
-description: |-
-  全栈开发工程师 - 与 `@fullstack-dev` 并行的第二实现轨（独立模块/API/页面岛）。PM 在 tasks 可并行或加速时指派；须写明模块边界与分支，勿当作闲置备用。
-  Fullstack Developer (Track 2) - the second implementation track parallel to `@fullstack-dev` (independent modules/APIs/page islands). PM should assign this role when tasks can run in parallel or when acceleration is needed, with explicit module boundaries and branch ownership.
+description: "Fullstack Developer (Track 2) - the second implementation track parallel to `@fullstack-dev` (independent modules/APIs/page islands). PM should assign this role when tasks can run in parallel or when acceleration is needed, with explicit module boundaries and branch ownership."
 mode: subagent
 tools:
   write: true

package/agents/fullstack-dev.md

@@ -1,8 +1,6 @@
 ---
 name: fullstack-dev
-description: |-
-  全栈开发工程师 - 后端主导的全栈实现（API/业务/数据层）。UI 重或新页面多时由 PM 拆给 `@frontend-dev`；第二并行实现轨用 `@fullstack-dev-2`。Hotfix/单流小改可一人完成。
-  Fullstack Developer - backend-led fullstack implementation (API/business/data). PM should split UI-heavy or new-page work to `@frontend-dev`, and use `@fullstack-dev-2` for the second parallel track. One developer can handle hotfixes or small single-stream updates.
+description: "Fullstack Developer - backend-led fullstack implementation (API/business/data). PM should split UI-heavy or new-page work to `@frontend-dev`, and use `@fullstack-dev-2` for the second parallel track. One developer can handle hotfixes or small single-stream updates."
 mode: subagent
 tools:
   write: true

package/agents/ops-engineer.md

@@ -1,8 +1,6 @@
 ---
 name: ops-engineer
-description: |-
-  运维工程师 - 部署、监控和基础设施。
-  Ops Engineer - deployment, monitoring, and infrastructure operations, including CI/CD and observability.
+description: "Ops Engineer - deployment, monitoring, and infrastructure operations, including CI/CD and observability."
 mode: subagent
 tools:
   write: true

package/agents/product-manager.md

@@ -1,8 +1,6 @@
 ---
 name: product-manager
-description: |-
-  产品经理 - 需求分析、产品规划、市场/用户研究与产品向文档编写。
-  Product Manager - requirements analysis, product planning, market/user research, and product-facing documentation.
+description: "Product Manager - requirements analysis, product planning, market/user research, and product-facing documentation."
 mode: subagent
 tools:
   write: true

package/agents/project-manager.md

@@ -1,8 +1,6 @@
 ---
 name: project-manager
-description: |-
-  项目经理 - 协调开发团队，管理项目进度。
-  Project Manager - coordinate the team, track progress, and orchestrate execution across roles.
+description: "Project Manager - coordinate the team, track progress, and orchestrate execution across roles."
 mode: primary
 tools:
   write: true

package/agents/prompt-engineer.md

@@ -1,8 +1,6 @@
 ---
 name: prompt-engineer
-description: |-
-  提示词工程师 - 设计与优化 Agent 提示词与技能。
-  Prompt Engineer - design and optimize prompts and skills for agents, including refactoring and debugging prompt systems.
+description: "Prompt Engineer - design and optimize prompts and skills for agents, including refactoring and debugging prompt systems."
 mode: subagent
 tools:
   write: true

package/agents/qa-engineer.md

@@ -1,8 +1,6 @@
 ---
 name: qa-engineer
-description: |-
-  测试工程师 - L4 验收席（分级派发）：DoD 映射、residual 验证、report-only 调查。
-  QA Engineer - tiered L4 acceptance (DoD mapping, residual verify, report-only); dispatched when PM sets QA gate mandatory.
+description: "QA Engineer - tiered L4 acceptance (DoD mapping, residual verify, report-only); dispatched when PM sets QA gate mandatory."
 mode: subagent
 tools:
   write: true

package/agents/qc-specialist-2.md

@@ -1,8 +1,6 @@
 ---
 name: qc-specialist-2
-description: |-
-  质量控制专家（Reviewer #2）- 代码审查（diff / 逻辑 / 安全正确性）。非测试执行席。
-  Quality Control Specialist (Reviewer #2) - code review (diff, logic, security/correctness). Not a test runner.
+description: "Quality Control Specialist (Reviewer #2) - code review (diff, logic, security/correctness). Not a test runner."
 mode: subagent
 tools:
   write: true

package/agents/qc-specialist-3.md

@@ -1,8 +1,6 @@
 ---
 name: qc-specialist-3
-description: |-
-  质量控制专家（Reviewer #3）- 代码审查（diff / 逻辑 / 性能可靠性）。非测试执行席。
-  Quality Control Specialist (Reviewer #3) - code review (diff, logic, performance/reliability). Not a test runner.
+description: "Quality Control Specialist (Reviewer #3) - code review (diff, logic, performance/reliability). Not a test runner."
 mode: subagent
 tools:
   write: true

package/agents/qc-specialist.md

@@ -1,8 +1,6 @@
 ---
 name: qc-specialist
-description: |-
-  质量控制专家（Reviewer #1）- 代码审查（diff / 逻辑 / 架构可维护性）。非测试执行席。
-  Quality Control Specialist (Reviewer #1) - code review (diff, logic, maintainability). Not a test runner.
+description: "Quality Control Specialist (Reviewer #1) - code review (diff, logic, maintainability). Not a test runner."
 mode: subagent
 tools:
   write: true

package/agents/writing-specialist.md

@@ -1,8 +1,6 @@
 ---
 name: writing-specialist
-description: |-
-  写作专家 - 文档写作、小说写作、文案写作与脚本写作。
-  Writing Specialist - documentation, fiction, copywriting, and script writing.
+description: "Writing Specialist - documentation, fiction, copywriting, and script writing."
 mode: subagent
 tools:
   write: true

package/harness-agents/architect.md

@@ -1,8 +1,6 @@
 ---
 name: architect
-description: |-
-  技术架构师 - 系统设计、技术决策与技术向文档编写（架构说明、ADR、接口契约等）。
-  Architect - system design, technical decisions, and technical documentation (architecture notes, ADRs, interface contracts).
+description: "Architect - system design, technical decisions, and technical documentation (architecture notes, ADRs, interface contracts)."
 mode: subagent
 tools:
   write: true

package/harness-agents/code-reviewer.md

@@ -1,8 +1,6 @@
 ---
 name: code-reviewer
-description: |-
-  代码审查员 - SDD 任务级审查（L2）+ 代码库审计执行（audit）。只读席位：不实现、不修代码、不占 QC 席。
-  Code Reviewer - SDD per-task review (L2) + codebase audit execution (audit). Read-only seat: does not implement, fix, or occupy a QC seat.
+description: "Code Reviewer - SDD per-task review (L2) + codebase audit execution (audit). Read-only seat: does not implement, fix, or occupy a QC seat."
 mode: subagent
 tools:
   write: true

package/harness-agents/frontend-dev.md

@@ -1,8 +1,6 @@
 ---
 name: frontend-dev
-description: |-
-  前端开发工程师 - 页面/组件/交互/a11y/前端性能。全栈功能里默认前端主责（与 `@fullstack-dev` 分轨）；纯 UI 任务首选本角色。
-  Frontend Developer - pages/components/interactions/accessibility/frontend performance. This is the default frontend owner in fullstack work (split with `@fullstack-dev`) and the preferred role for pure UI tasks.
+description: "Frontend Developer - pages/components/interactions/accessibility/frontend performance. This is the default frontend owner in fullstack work (split with `@fullstack-dev`) and the preferred role for pure UI tasks."
 mode: subagent
 tools:
   write: true

package/harness-agents/fullstack-dev-2.md

@@ -1,8 +1,6 @@
 ---
 name: fullstack-dev-2
-description: |-
-  全栈开发工程师 - 与 `@fullstack-dev` 并行的第二实现轨（独立模块/API/页面岛）。PM 在 tasks 可并行或加速时指派；须写明模块边界与分支，勿当作闲置备用。
-  Fullstack Developer (Track 2) - the second implementation track parallel to `@fullstack-dev` (independent modules/APIs/page islands). PM should assign this role when tasks can run in parallel or when acceleration is needed, with explicit module boundaries and branch ownership.
+description: "Fullstack Developer (Track 2) - the second implementation track parallel to `@fullstack-dev` (independent modules/APIs/page islands). PM should assign this role when tasks can run in parallel or when acceleration is needed, with explicit module boundaries and branch ownership."
 mode: subagent
 tools:
   write: true

package/harness-agents/fullstack-dev.md

@@ -1,8 +1,6 @@
 ---
 name: fullstack-dev
-description: |-
-  全栈开发工程师 - 后端主导的全栈实现（API/业务/数据层）。UI 重或新页面多时由 PM 拆给 `@frontend-dev`；第二并行实现轨用 `@fullstack-dev-2`。Hotfix/单流小改可一人完成。
-  Fullstack Developer - backend-led fullstack implementation (API/business/data). PM should split UI-heavy or new-page work to `@frontend-dev`, and use `@fullstack-dev-2` for the second parallel track. One developer can handle hotfixes or small single-stream updates.
+description: "Fullstack Developer - backend-led fullstack implementation (API/business/data). PM should split UI-heavy or new-page work to `@frontend-dev`, and use `@fullstack-dev-2` for the second parallel track. One developer can handle hotfixes or small single-stream updates."
 mode: subagent
 tools:
   write: true

package/harness-agents/ops-engineer.md

@@ -1,8 +1,6 @@
 ---
 name: ops-engineer
-description: |-
-  运维工程师 - 部署、监控和基础设施。
-  Ops Engineer - deployment, monitoring, and infrastructure operations, including CI/CD and observability.
+description: "Ops Engineer - deployment, monitoring, and infrastructure operations, including CI/CD and observability."
 mode: subagent
 tools:
   write: true

package/harness-agents/product-manager.md

@@ -1,8 +1,6 @@
 ---
 name: product-manager
-description: |-
-  产品经理 - 需求分析、产品规划、市场/用户研究与产品向文档编写。
-  Product Manager - requirements analysis, product planning, market/user research, and product-facing documentation.
+description: "Product Manager - requirements analysis, product planning, market/user research, and product-facing documentation."
 mode: subagent
 tools:
   write: true

package/harness-agents/project-manager.md

@@ -1,8 +1,6 @@
 ---
 name: project-manager
-description: |-
-  项目经理 - 协调开发团队，管理项目进度。
-  Project Manager - coordinate the team, track progress, and orchestrate execution across roles.
+description: "Project Manager - coordinate the team, track progress, and orchestrate execution across roles."
 mode: primary
 tools:
   write: true

package/harness-agents/prompt-engineer.md

@@ -1,8 +1,6 @@
 ---
 name: prompt-engineer
-description: |-
-  提示词工程师 - 设计与优化 Agent 提示词与技能。
-  Prompt Engineer - design and optimize prompts and skills for agents, including refactoring and debugging prompt systems.
+description: "Prompt Engineer - design and optimize prompts and skills for agents, including refactoring and debugging prompt systems."
 mode: subagent
 tools:
   write: true

package/harness-agents/qa-engineer.md

@@ -1,8 +1,6 @@
 ---
 name: qa-engineer
-description: |-
-  测试工程师 - L4 验收席（分级派发）：DoD 映射、residual 验证、report-only 调查。
-  QA Engineer - tiered L4 acceptance (DoD mapping, residual verify, report-only); dispatched when PM sets QA gate mandatory.
+description: "QA Engineer - tiered L4 acceptance (DoD mapping, residual verify, report-only); dispatched when PM sets QA gate mandatory."
 mode: subagent
 tools:
   write: true

package/harness-agents/qc-specialist-2.md

@@ -1,8 +1,6 @@
 ---
 name: qc-specialist-2
-description: |-
-  质量控制专家（Reviewer #2）- 代码审查（diff / 逻辑 / 安全正确性）。非测试执行席。
-  Quality Control Specialist (Reviewer #2) - code review (diff, logic, security/correctness). Not a test runner.
+description: "Quality Control Specialist (Reviewer #2) - code review (diff, logic, security/correctness). Not a test runner."
 mode: subagent
 tools:
   write: true

package/harness-agents/qc-specialist-3.md

@@ -1,8 +1,6 @@
 ---
 name: qc-specialist-3
-description: |-
-  质量控制专家（Reviewer #3）- 代码审查（diff / 逻辑 / 性能可靠性）。非测试执行席。
-  Quality Control Specialist (Reviewer #3) - code review (diff, logic, performance/reliability). Not a test runner.
+description: "Quality Control Specialist (Reviewer #3) - code review (diff, logic, performance/reliability). Not a test runner."
 mode: subagent
 tools:
   write: true

package/harness-agents/qc-specialist.md

@@ -1,8 +1,6 @@
 ---
 name: qc-specialist
-description: |-
-  质量控制专家（Reviewer #1）- 代码审查（diff / 逻辑 / 架构可维护性）。非测试执行席。
-  Quality Control Specialist (Reviewer #1) - code review (diff, logic, maintainability). Not a test runner.
+description: "Quality Control Specialist (Reviewer #1) - code review (diff, logic, maintainability). Not a test runner."
 mode: subagent
 tools:
   write: true

package/harness-agents/writing-specialist.md

@@ -1,8 +1,6 @@
 ---
 name: writing-specialist
-description: |-
-  写作专家 - 文档写作、小说写作、文案写作与脚本写作。
-  Writing Specialist - documentation, fiction, copywriting, and script writing.
+description: "Writing Specialist - documentation, fiction, copywriting, and script writing."
 mode: subagent
 tools:
   write: true

package/harness-skills/mstar-host/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: mstar-host
-description: Morning Star host adapter (OpenCode, Cursor, Codex, Kimi, ZCode, omp, dsh). Use after mstar-harness-core whenever host entry, clarify, dispatch, or plan UX differs by platform - OpenCode question/task-tool subagent invoke, Cursor /pm and CreatePlan/SwitchMode dual-write and Task parallel QC, Codex plugin skills plus Plan/Goal Mode, Kimi Agent/AgentSwarm with built-in subagent types only (coder/explore/plan) and role-in-prompt binding, ZCode Agent/AskUserQuestion/EnterPlanMode with built-in subagent types and role-in-prompt binding, omp task/ask/hub preferring live-schema role agents (agents/*.md) with C5b skill-load binding (generic task/scout only as fallback), dsh (DeepSeek Harness) subagent tool with in-process engine gates and bundled mstar commands, sandboxed tools, and tool discovery. Auto-detect host from session tools; then Read references/<host>.md. Always load after mstar-harness-core.
+description: Morning Star host adapter for OpenCode, Cursor, Codex, Kimi, ZCode, omp, and dsh. Use after mstar-harness-core whenever host entry, clarify, dispatch, plan mode, or tool UX differs by platform - per-host subagent invoke, plan-mode bridge, parallel dispatch, and skill/command loading. Auto-detect the host from session tools, then Read references/<host>.md. Always load after mstar-harness-core.
 ---
 
 # Morning Star Host Adapter

package/harness-skills/mstar-host/references/_shared/host-role-binding-core.md

@@ -1,21 +1,21 @@
 # Host role-binding core (C5/C5b)
 
-Shared role-binding contract for **Kimi / ZCode** — hosts where Morning Star role ids are **never** valid invoke types, so picking `subagent_type` alone cannot carry Assignment IDENTITY + skill load.
+Shared role-binding contract for **Kimi** — the host where Morning Star role ids are **never** valid invoke types, so picking `subagent_type` alone cannot carry Assignment IDENTITY + skill load.
 
-Other hosts follow their own `references/<host>.md` and do **not** use this file for agent selection. Read this file only when the active host reference is Kimi or ZCode.
+Other hosts follow their own `references/<host>.md` and do **not** use this file for agent selection. Read this file only when the active host reference is Kimi.
 
-## C5 — invoke type selection (Kimi / ZCode)
+## C5 — invoke type selection (Kimi)
 
 | Host | Invoke-type selection (C5) | Prompt role binding (C5b) |
 |------|----------------------------|---------------------------|
-| **Kimi / ZCode** | Built-in types only (`coder` / `explore` / `plan`, …). Morning Star role ids are **never** valid invoke types. | **Required** |
+| **Kimi** | Built-in types only (`coder` / `explore` / `plan`, …). Morning Star role ids are **never** valid invoke types. | **Required** |
 
 - **Never invent** agent / subagent names. The **live tool schema** (plus the active host reference C5 table) is SSOT every session.
 - Morning Star role ids (`product-manager`, `fullstack-dev`, `qc-specialist`, …) are **not** valid invoke types — map via the host reference table (`coder` / `explore` / `plan`, …), then rely on C5b.
 
 ## C5b — role binding in prompt (required)
 
-Every dispatch on Kimi / ZCode **must** carry the played Morning Star role in the **Assignment** and in the **invoke prompt**:
+Every dispatch on Kimi **must** carry the played Morning Star role in the **Assignment** and in the **invoke prompt**:
 
 1. **`Execute as: <role-id>`** in Assignment (harness routing SSOT).
 2. **`Act as <role-id>`** (or equivalent) at the top of the invoke prompt.
@@ -48,4 +48,4 @@ Load: `mstar-harness-core` → `mstar-host` → `<host>.md` → `mstar-roles`
 <task body>
 ```
 
-The active host reference shows the matching invoke shape (same turn) — e.g. Kimi/ZCode `Agent` with `subagent_type: "coder"` plus this C5b body.
+The active host reference shows the matching invoke shape (same turn) — e.g. Kimi `Agent` with `subagent_type: "coder"` plus this C5b body.

package/harness-skills/mstar-host/references/omp.md

@@ -106,7 +106,7 @@ Single-task shorthand may exist depending on host version — always match the l
 
 ### Role binding in prompt (C5b — required)
 
-omp C5/C5b SSOT is **this file** — do **not** use `_shared/host-role-binding-core.md` (that file is Kimi/ZCode only). Even when `agent` already matches the role id, still bind Morning Star process in the Assignment / `task` body (agent shell ≠ full role prompt; skill load is not automatic).
+omp C5/C5b SSOT is **this file**. Even when `agent` already matches the role id, still bind Morning Star process in the Assignment / `task` body (agent shell ≠ full role prompt; skill load is not automatic).
 
 Required in every role dispatch:
 

package/harness-skills/mstar-host/references/zcode.md

@@ -11,9 +11,9 @@ Parallel PM dispatch: read **`parallel-dispatch.md`** when dispatching **N ≥ 2
 - Plugin manifest: **`.zcode-plugin/plugin.json`** (plugin root is the **repo root**; paths stay `./skills/`, `./commands/`, `./agents/`).
 - Runtime skills: repo `skills/` mounted by the plugin (`"skills": "./skills/"`).
 - Plugin commands: repo `commands/` → `/morning-star-harness:<name>` (e.g. `/morning-star-harness:iteration-start`).
-- Plugin agents: repo `agents/*.md` (role frontmatter `name` + `description`); ZCode reads them as subagent definitions but **does not** expose custom named `subagent_type` values for Morning Star roles (see C5).
+- Plugin agents: repo `agents/*.md` (role frontmatter `name` + **single-line quoted English `description`** — ZCode's flat frontmatter parser cannot read `|-` block scalars and silently drops nested `tools`/`permission` maps). ZCode registers them as callable `subagent_type`s in both bare (`fullstack-dev`) and plugin-qualified (`morning-star-harness:fullstack-dev`) forms (see C5).
 - **No `sessionStart.skill`** (ZCode has no Kimi-style session auto-load) — enter PM manually via **`/morning-star-harness:pm`** or the **`pm`** skill, then **Read next** → `mstar-harness-core` → `project-manager.md`.
-- Install (user-scoped, recommended): `npx @mstar-harness/cli init --target zcode --scope global`, then in ZCode **Settings → Plugin Management → Discover** install **morning-star-harness** from the **mstar-local** marketplace (or add `github:btspoony/mstar-harness` as a marketplace directly).
+- Install (user-scoped, recommended): `npx @mstar-harness/cli init --target zcode --scope global`, then in ZCode **Settings → Plugin Management → Discover** install **morning-star-harness** from the **mstar-local** marketplace. Direct-add also works without the CLI: the repo ships `.claude-plugin/marketplace.json` (root `marketplace.json` fallback), which ZCode discovers when refreshing a `github`-source marketplace.
 - Plugins are **user-scoped** (all projects); managed copy lives under `~/.zcode/cli/plugins/` after install.
 - Project `.agents/skills/` symlinks are **not** required when using the plugin — commands and skills come from the plugin mount.
 
@@ -31,7 +31,7 @@ Use skill names in prompts and references. Avoid absolute local paths unless mai
 
 | ZCode tool | Harness use |
 |------------|-------------|
-| **Agent** | Primary dispatch — delegate one subagent task (`subagent_type`: built-in profiles such as `general-purpose` / `Explore`) |
+| **Agent** | Primary dispatch — delegate one subagent task (`subagent_type`: bare Morning Star role id, e.g. `fullstack-dev`; `general-purpose` fallback; `Explore` read-only) |
 | **AskUserQuestion** | Structured clarify (1–4 questions, 2–4 options each); prefer over free-form when choices are known |
 | **EnterPlanMode** / **ExitPlanMode** | Plan mode entry/approval → **`zcode-plan-mode-bridge.md`** |
 | **TodoWrite** | Session UX only; mirror to SSOT plan / workflow snapshot (`{WORKFLOW_DIR}/<id>/snapshot.json`) when durable |
@@ -40,34 +40,62 @@ Use skill names in prompts and references. Avoid absolute local paths unless mai
 | **Edit** / **Write** | Edits |
 | **Glob** / **Grep** | Search (prefer over shell find/grep) |
 | **WebSearch** / **WebFetch** | External docs / facts |
+| **Skill** | Invoke bundled skills (`/morning-star-harness:pm`, `mstar-harness-core`, …) |
 | **TaskOutput** / **TaskStop** | Long-running task management when present |
 
 OpenCode-style `question`/`task`, Cursor **Task**, and Kimi **AgentSwarm** are **not** ZCode tools — do not assume them.
 
-## Role agents (C5 — hard constraint)
+## Role agents (C5)
 
-ZCode ships **built-in subagent types only** (e.g. `general-purpose`, `Explore`). Valid **`subagent_type`** values are the host's built-in profiles:
+ZCode registers the plugin's Morning Star agents (`agents/*.md`) as callable **`subagent_type`** values alongside the built-in profiles, in both **bare role id** (`fullstack-dev`) and **plugin-qualified** (`morning-star-harness:fullstack-dev`) forms. Prefer the **bare role id** (verified on ZCode 0.16.5: the bare form loads the role shell body; the qualified form can silently fall back to the default prompt):
 
 | `subagent_type` | ZCode profile | Harness mapping |
 |-----------------|---------------|-----------------|
-| `Explore` | Read-only exploration | Orientation, codebase survey, Prepare explore passes |
-| `general-purpose` | General implementation / research | **All other Morning Star roles** (`product-manager`, `fullstack-dev`, `qc-specialist`, …) |
+| `<role-id>` (bare, e.g. `fullstack-dev`, `qc-specialist`) | Registered plugin agent | **Preferred** for the matching Morning Star role |
+| `Explore` | Built-in read-only exploration | Orientation, codebase survey, Prepare explore passes |
+| `general-purpose` | Built-in general profile | **Universal fallback** when a role id is not exposed |
 
-Morning Star role ids (`project-manager`, `fullstack-dev`, `qc-specialist`, …) are **not** valid `subagent_type` values. Although `agents/*.md` are read by ZCode, the host does not register them as callable named agent types the way Codex TOML or Cursor `subagent_type` role ids do.
+Role shells are thin — each body only states the role identity and points at the `mstar-roles` skill — so C5b prompt binding stays **required** on every dispatch regardless of `subagent_type`.
 
 ### Role binding in prompt (C5b — required)
 
-Role-binding contract + Assignment template → **`_shared/host-role-binding-core.md`** (C5/C5b). ZCode-specific invoke shapes, same turn:
+ZCode C5/C5b SSOT is **this file** — do **not** load `_shared/host-role-binding-core.md` (that file is Kimi only). Role shells are thin (identity + skill pointer, not the full role prompt), so every role dispatch must carry:
+
+1. **`Execute as: <role-id>`** in Assignment (harness routing SSOT).
+2. **`Act as <role-id>`** (or equivalent) at the top of the invoke prompt.
+3. **Skill load list** — instruct the subagent to read `mstar-roles` → `references/<role-id>.md` (or shared reference + parameters) and topic skills per that reference.
+4. **`subagent_type`** — bare Morning Star role id per C5; `general-purpose` fallback.
+
+Paste-only Assignment **without** an invoke call is **not** dispatch. Anti-recursion NEVER: leaf executors are already `Execute as` — no recursive invoke of the same role; Assignment wins (`Delegation: forbidden` unless stated). **Never** multiple implementer invokes in one message for the same plan (SDD serial → **`parallel-dispatch.md`** § SDD implement).
+
+ZCode invoke shape (same turn):
 
 ```text
 Agent(
-  subagent_type: "general-purpose",
+  subagent_type: "fullstack-dev",
   description: "<short task label>",
-  prompt: "<full Assignment body including Act as + skill load>"
+  prompt: "<full Assignment body including Execute as + Act as + skill load>"
 )
 ```
 
-For **`Explore`** orientation:
+Assignment / prompt template:
+
+```markdown
+## Assignment
+
+**Execute as**: fullstack-dev
+**Delegation**: forbidden
+**Working branch**: feat/example
+**Plan Path**: .mstar/plans/20260717-example.md
+
+**IDENTITY:** You ARE `fullstack-dev`. Act as `fullstack-dev` for this task.
+
+Load: `mstar-harness-core` → `mstar-host` → `zcode.md` → `mstar-roles` → `references/fullstack-dev-shared.md` → topic skills per that reference.
+
+<task body>
+```
+
+Role id not exposed (older ZCode) → `subagent_type: "general-purpose"` with the same body. For **`Explore`** orientation:
 
 ```text
 Agent(subagent_type: "Explore", description: "...", prompt: "... Act as explore-only orientation; Execute as: n/a ...")
@@ -75,24 +103,24 @@ Agent(subagent_type: "Explore", description: "...", prompt: "... Act as explore-
 
 ## PM dispatch (`Agent`)
 
-Harness **dispatch** on ZCode = **one or more `Agent` tool calls** with correct **`subagent_type`** and role-bound prompts (C5b → **`_shared/host-role-binding-core.md`**). N-parallel / 1-Assignment-1-invoke / paste-only mechanics → **`parallel-dispatch.md`**.
+Harness **dispatch** on ZCode = **one or more `Agent` tool calls** with correct **`subagent_type`** and role-bound prompts (C5b — this file). N-parallel / 1-Assignment-1-invoke / paste-only mechanics → **`parallel-dispatch.md`**.
 
 | Harness | ZCode |
 |---------|-------|
 | `Execute as: <role-id>` | Role id in Assignment + **Act as** + skill load in **Agent** prompt (C5b) |
-| `subagent_type` for invoke | built-in profiles only (typically `general-purpose`; `Explore` for read-only) |
+| `subagent_type` for invoke | bare role id (e.g. `fullstack-dev`, `qc-specialist`); `general-purpose` fallback; `Explore` for read-only |
 | Parallel batch **N** | **N `Agent`** calls in **one assistant message** |
 
 ### QC default
 
-- **`Execution mode: sdd`**: **N=3** `Agent` calls (`qc-specialist`, `qc-specialist-2`, `qc-specialist-3`) — each prompt **Act as** the respective QC role, all `subagent_type: "general-purpose"` (N rules → `parallel-dispatch.md`).
+- **`Execution mode: sdd`**: **N=3** `Agent` calls (`qc-specialist`, `qc-specialist-2`, `qc-specialist-3`) — each prompt **Act as** the respective QC role, all via bare `subagent_type: "qc-specialist"` / `"qc-specialist-2"` / `"qc-specialist-3"` (`general-purpose` fallback; N rules → `parallel-dispatch.md`).
 - **`inline`**: **N=1** per `parallel-dispatch.md`.
 
 Cannot emit required **N** → **`Blocked`**.
 
 ### SDD implement (serial)
 
-- **`Execution mode: sdd`**: one implementer **`Agent`** per task id; task reviewer = new **`Agent`** with **Act as `code-reviewer`** (ZCode L2 review; not qc-specialist*), always via generic fallback `subagent_type: "general-purpose"` per C5 — no sticky resume unless host adds it later. Serial rule → **`parallel-dispatch.md`** § SDD implement.
+- **`Execution mode: sdd`**: one implementer **`Agent`** per task id (bare role id per C5, `general-purpose` fallback); task reviewer = new **`Agent`** with **Act as `code-reviewer`** (`subagent_type: "code-reviewer"`, `general-purpose` fallback; ZCode L2 review; not qc-specialist*), always with C5b prompt binding — no sticky resume unless host adds it later. Serial rule → **`parallel-dispatch.md`** § SDD implement.
 - **Never** multiple implementer Agents in one message for the same plan.
 
 ## Clarify
@@ -125,5 +153,5 @@ Cannot emit required **N** → **`Blocked`**.
 
 - Plugin install materializes source under `~/.zcode/cli/plugins/` — edit the harness checkout + reinstall to pick up harness changes.
 - Session todos (`TodoWrite`) are not durable SSOT unless mirrored to `{HARNESS_DIR}`.
-- No custom ZCode agent profiles for Morning Star roles — role binding is **always** prompt + skill load (C5b).
+- Morning Star roles **are** registered as ZCode subagent types from `agents/*.md` — but the shells are thin, so role binding is **still always** prompt + skill load (C5b); prefer bare role ids, the qualified form can fall back to the default prompt.
 - ZCode has no `sessionStart.skill`; new sessions do **not** auto-load PM — invoke `/morning-star-harness:pm` or `/skill:pm` manually.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mstar-harness/omp",
-  "version": "3.6.0-alpha.4",
+  "version": "3.6.1",
   "description": "Morning Star harness omp (Oh My Pi) plugin — engine-backed runtime gates (status/dispatch/lease) + mstar-* skills, commands, and agents for omp.",
   "license": "MIT",
   "repository": {

package/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "morning-star-harness",
-  "version": "3.6.0-alpha.4",
+  "version": "3.6.1",
   "description": "Morning Star — Agent Plugin for harness engineering workflows: a TS Harness Workflow Engine (@mstar-harness/engine) that enforces deterministic workflow gates, plus mstar-* judgment skills for multi-agent code delivery across OpenCode, Cursor, Codex, Kimi Code, ZCode, and omp.",
   "author": {
     "name": "Bohao Tang",

package/skills/mstar-host/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: mstar-host
-description: Morning Star host adapter (OpenCode, Cursor, Codex, Kimi, ZCode, omp, dsh). Use after mstar-harness-core whenever host entry, clarify, dispatch, or plan UX differs by platform - OpenCode question/task-tool subagent invoke, Cursor /pm and CreatePlan/SwitchMode dual-write and Task parallel QC, Codex plugin skills plus Plan/Goal Mode, Kimi Agent/AgentSwarm with built-in subagent types only (coder/explore/plan) and role-in-prompt binding, ZCode Agent/AskUserQuestion/EnterPlanMode with built-in subagent types and role-in-prompt binding, omp task/ask/hub preferring live-schema role agents (agents/*.md) with C5b skill-load binding (generic task/scout only as fallback), dsh (DeepSeek Harness) subagent tool with in-process engine gates and bundled mstar commands, sandboxed tools, and tool discovery. Auto-detect host from session tools; then Read references/<host>.md. Always load after mstar-harness-core.
+description: Morning Star host adapter for OpenCode, Cursor, Codex, Kimi, ZCode, omp, and dsh. Use after mstar-harness-core whenever host entry, clarify, dispatch, plan mode, or tool UX differs by platform - per-host subagent invoke, plan-mode bridge, parallel dispatch, and skill/command loading. Auto-detect the host from session tools, then Read references/<host>.md. Always load after mstar-harness-core.
 ---
 
 # Morning Star Host Adapter

package/skills/mstar-host/references/_shared/host-role-binding-core.md

@@ -1,21 +1,21 @@
 # Host role-binding core (C5/C5b)
 
-Shared role-binding contract for **Kimi / ZCode** — hosts where Morning Star role ids are **never** valid invoke types, so picking `subagent_type` alone cannot carry Assignment IDENTITY + skill load.
+Shared role-binding contract for **Kimi** — the host where Morning Star role ids are **never** valid invoke types, so picking `subagent_type` alone cannot carry Assignment IDENTITY + skill load.
 
-Other hosts follow their own `references/<host>.md` and do **not** use this file for agent selection. Read this file only when the active host reference is Kimi or ZCode.
+Other hosts follow their own `references/<host>.md` and do **not** use this file for agent selection. Read this file only when the active host reference is Kimi.
 
-## C5 — invoke type selection (Kimi / ZCode)
+## C5 — invoke type selection (Kimi)
 
 | Host | Invoke-type selection (C5) | Prompt role binding (C5b) |
 |------|----------------------------|---------------------------|
-| **Kimi / ZCode** | Built-in types only (`coder` / `explore` / `plan`, …). Morning Star role ids are **never** valid invoke types. | **Required** |
+| **Kimi** | Built-in types only (`coder` / `explore` / `plan`, …). Morning Star role ids are **never** valid invoke types. | **Required** |
 
 - **Never invent** agent / subagent names. The **live tool schema** (plus the active host reference C5 table) is SSOT every session.
 - Morning Star role ids (`product-manager`, `fullstack-dev`, `qc-specialist`, …) are **not** valid invoke types — map via the host reference table (`coder` / `explore` / `plan`, …), then rely on C5b.
 
 ## C5b — role binding in prompt (required)
 
-Every dispatch on Kimi / ZCode **must** carry the played Morning Star role in the **Assignment** and in the **invoke prompt**:
+Every dispatch on Kimi **must** carry the played Morning Star role in the **Assignment** and in the **invoke prompt**:
 
 1. **`Execute as: <role-id>`** in Assignment (harness routing SSOT).
 2. **`Act as <role-id>`** (or equivalent) at the top of the invoke prompt.
@@ -48,4 +48,4 @@ Load: `mstar-harness-core` → `mstar-host` → `<host>.md` → `mstar-roles`
 <task body>
 ```
 
-The active host reference shows the matching invoke shape (same turn) — e.g. Kimi/ZCode `Agent` with `subagent_type: "coder"` plus this C5b body.
+The active host reference shows the matching invoke shape (same turn) — e.g. Kimi `Agent` with `subagent_type: "coder"` plus this C5b body.

package/skills/mstar-host/references/omp.md

@@ -106,7 +106,7 @@ Single-task shorthand may exist depending on host version — always match the l
 
 ### Role binding in prompt (C5b — required)
 
-omp C5/C5b SSOT is **this file** — do **not** use `_shared/host-role-binding-core.md` (that file is Kimi/ZCode only). Even when `agent` already matches the role id, still bind Morning Star process in the Assignment / `task` body (agent shell ≠ full role prompt; skill load is not automatic).
+omp C5/C5b SSOT is **this file**. Even when `agent` already matches the role id, still bind Morning Star process in the Assignment / `task` body (agent shell ≠ full role prompt; skill load is not automatic).
 
 Required in every role dispatch:
 

package/skills/mstar-host/references/zcode.md

@@ -11,9 +11,9 @@ Parallel PM dispatch: read **`parallel-dispatch.md`** when dispatching **N ≥ 2
 - Plugin manifest: **`.zcode-plugin/plugin.json`** (plugin root is the **repo root**; paths stay `./skills/`, `./commands/`, `./agents/`).
 - Runtime skills: repo `skills/` mounted by the plugin (`"skills": "./skills/"`).
 - Plugin commands: repo `commands/` → `/morning-star-harness:<name>` (e.g. `/morning-star-harness:iteration-start`).
-- Plugin agents: repo `agents/*.md` (role frontmatter `name` + `description`); ZCode reads them as subagent definitions but **does not** expose custom named `subagent_type` values for Morning Star roles (see C5).
+- Plugin agents: repo `agents/*.md` (role frontmatter `name` + **single-line quoted English `description`** — ZCode's flat frontmatter parser cannot read `|-` block scalars and silently drops nested `tools`/`permission` maps). ZCode registers them as callable `subagent_type`s in both bare (`fullstack-dev`) and plugin-qualified (`morning-star-harness:fullstack-dev`) forms (see C5).
 - **No `sessionStart.skill`** (ZCode has no Kimi-style session auto-load) — enter PM manually via **`/morning-star-harness:pm`** or the **`pm`** skill, then **Read next** → `mstar-harness-core` → `project-manager.md`.
-- Install (user-scoped, recommended): `npx @mstar-harness/cli init --target zcode --scope global`, then in ZCode **Settings → Plugin Management → Discover** install **morning-star-harness** from the **mstar-local** marketplace (or add `github:btspoony/mstar-harness` as a marketplace directly).
+- Install (user-scoped, recommended): `npx @mstar-harness/cli init --target zcode --scope global`, then in ZCode **Settings → Plugin Management → Discover** install **morning-star-harness** from the **mstar-local** marketplace. Direct-add also works without the CLI: the repo ships `.claude-plugin/marketplace.json` (root `marketplace.json` fallback), which ZCode discovers when refreshing a `github`-source marketplace.
 - Plugins are **user-scoped** (all projects); managed copy lives under `~/.zcode/cli/plugins/` after install.
 - Project `.agents/skills/` symlinks are **not** required when using the plugin — commands and skills come from the plugin mount.
 
@@ -31,7 +31,7 @@ Use skill names in prompts and references. Avoid absolute local paths unless mai
 
 | ZCode tool | Harness use |
 |------------|-------------|
-| **Agent** | Primary dispatch — delegate one subagent task (`subagent_type`: built-in profiles such as `general-purpose` / `Explore`) |
+| **Agent** | Primary dispatch — delegate one subagent task (`subagent_type`: bare Morning Star role id, e.g. `fullstack-dev`; `general-purpose` fallback; `Explore` read-only) |
 | **AskUserQuestion** | Structured clarify (1–4 questions, 2–4 options each); prefer over free-form when choices are known |
 | **EnterPlanMode** / **ExitPlanMode** | Plan mode entry/approval → **`zcode-plan-mode-bridge.md`** |
 | **TodoWrite** | Session UX only; mirror to SSOT plan / workflow snapshot (`{WORKFLOW_DIR}/<id>/snapshot.json`) when durable |
@@ -40,34 +40,62 @@ Use skill names in prompts and references. Avoid absolute local paths unless mai
 | **Edit** / **Write** | Edits |
 | **Glob** / **Grep** | Search (prefer over shell find/grep) |
 | **WebSearch** / **WebFetch** | External docs / facts |
+| **Skill** | Invoke bundled skills (`/morning-star-harness:pm`, `mstar-harness-core`, …) |
 | **TaskOutput** / **TaskStop** | Long-running task management when present |
 
 OpenCode-style `question`/`task`, Cursor **Task**, and Kimi **AgentSwarm** are **not** ZCode tools — do not assume them.
 
-## Role agents (C5 — hard constraint)
+## Role agents (C5)
 
-ZCode ships **built-in subagent types only** (e.g. `general-purpose`, `Explore`). Valid **`subagent_type`** values are the host's built-in profiles:
+ZCode registers the plugin's Morning Star agents (`agents/*.md`) as callable **`subagent_type`** values alongside the built-in profiles, in both **bare role id** (`fullstack-dev`) and **plugin-qualified** (`morning-star-harness:fullstack-dev`) forms. Prefer the **bare role id** (verified on ZCode 0.16.5: the bare form loads the role shell body; the qualified form can silently fall back to the default prompt):
 
 | `subagent_type` | ZCode profile | Harness mapping |
 |-----------------|---------------|-----------------|
-| `Explore` | Read-only exploration | Orientation, codebase survey, Prepare explore passes |
-| `general-purpose` | General implementation / research | **All other Morning Star roles** (`product-manager`, `fullstack-dev`, `qc-specialist`, …) |
+| `<role-id>` (bare, e.g. `fullstack-dev`, `qc-specialist`) | Registered plugin agent | **Preferred** for the matching Morning Star role |
+| `Explore` | Built-in read-only exploration | Orientation, codebase survey, Prepare explore passes |
+| `general-purpose` | Built-in general profile | **Universal fallback** when a role id is not exposed |
 
-Morning Star role ids (`project-manager`, `fullstack-dev`, `qc-specialist`, …) are **not** valid `subagent_type` values. Although `agents/*.md` are read by ZCode, the host does not register them as callable named agent types the way Codex TOML or Cursor `subagent_type` role ids do.
+Role shells are thin — each body only states the role identity and points at the `mstar-roles` skill — so C5b prompt binding stays **required** on every dispatch regardless of `subagent_type`.
 
 ### Role binding in prompt (C5b — required)
 
-Role-binding contract + Assignment template → **`_shared/host-role-binding-core.md`** (C5/C5b). ZCode-specific invoke shapes, same turn:
+ZCode C5/C5b SSOT is **this file** — do **not** load `_shared/host-role-binding-core.md` (that file is Kimi only). Role shells are thin (identity + skill pointer, not the full role prompt), so every role dispatch must carry:
+
+1. **`Execute as: <role-id>`** in Assignment (harness routing SSOT).
+2. **`Act as <role-id>`** (or equivalent) at the top of the invoke prompt.
+3. **Skill load list** — instruct the subagent to read `mstar-roles` → `references/<role-id>.md` (or shared reference + parameters) and topic skills per that reference.
+4. **`subagent_type`** — bare Morning Star role id per C5; `general-purpose` fallback.
+
+Paste-only Assignment **without** an invoke call is **not** dispatch. Anti-recursion NEVER: leaf executors are already `Execute as` — no recursive invoke of the same role; Assignment wins (`Delegation: forbidden` unless stated). **Never** multiple implementer invokes in one message for the same plan (SDD serial → **`parallel-dispatch.md`** § SDD implement).
+
+ZCode invoke shape (same turn):
 
 ```text
 Agent(
-  subagent_type: "general-purpose",
+  subagent_type: "fullstack-dev",
   description: "<short task label>",
-  prompt: "<full Assignment body including Act as + skill load>"
+  prompt: "<full Assignment body including Execute as + Act as + skill load>"
 )
 ```
 
-For **`Explore`** orientation:
+Assignment / prompt template:
+
+```markdown
+## Assignment
+
+**Execute as**: fullstack-dev
+**Delegation**: forbidden
+**Working branch**: feat/example
+**Plan Path**: .mstar/plans/20260717-example.md
+
+**IDENTITY:** You ARE `fullstack-dev`. Act as `fullstack-dev` for this task.
+
+Load: `mstar-harness-core` → `mstar-host` → `zcode.md` → `mstar-roles` → `references/fullstack-dev-shared.md` → topic skills per that reference.
+
+<task body>
+```
+
+Role id not exposed (older ZCode) → `subagent_type: "general-purpose"` with the same body. For **`Explore`** orientation:
 
 ```text
 Agent(subagent_type: "Explore", description: "...", prompt: "... Act as explore-only orientation; Execute as: n/a ...")
@@ -75,24 +103,24 @@ Agent(subagent_type: "Explore", description: "...", prompt: "... Act as explore-
 
 ## PM dispatch (`Agent`)
 
-Harness **dispatch** on ZCode = **one or more `Agent` tool calls** with correct **`subagent_type`** and role-bound prompts (C5b → **`_shared/host-role-binding-core.md`**). N-parallel / 1-Assignment-1-invoke / paste-only mechanics → **`parallel-dispatch.md`**.
+Harness **dispatch** on ZCode = **one or more `Agent` tool calls** with correct **`subagent_type`** and role-bound prompts (C5b — this file). N-parallel / 1-Assignment-1-invoke / paste-only mechanics → **`parallel-dispatch.md`**.
 
 | Harness | ZCode |
 |---------|-------|
 | `Execute as: <role-id>` | Role id in Assignment + **Act as** + skill load in **Agent** prompt (C5b) |
-| `subagent_type` for invoke | built-in profiles only (typically `general-purpose`; `Explore` for read-only) |
+| `subagent_type` for invoke | bare role id (e.g. `fullstack-dev`, `qc-specialist`); `general-purpose` fallback; `Explore` for read-only |
 | Parallel batch **N** | **N `Agent`** calls in **one assistant message** |
 
 ### QC default
 
-- **`Execution mode: sdd`**: **N=3** `Agent` calls (`qc-specialist`, `qc-specialist-2`, `qc-specialist-3`) — each prompt **Act as** the respective QC role, all `subagent_type: "general-purpose"` (N rules → `parallel-dispatch.md`).
+- **`Execution mode: sdd`**: **N=3** `Agent` calls (`qc-specialist`, `qc-specialist-2`, `qc-specialist-3`) — each prompt **Act as** the respective QC role, all via bare `subagent_type: "qc-specialist"` / `"qc-specialist-2"` / `"qc-specialist-3"` (`general-purpose` fallback; N rules → `parallel-dispatch.md`).
 - **`inline`**: **N=1** per `parallel-dispatch.md`.
 
 Cannot emit required **N** → **`Blocked`**.
 
 ### SDD implement (serial)
 
-- **`Execution mode: sdd`**: one implementer **`Agent`** per task id; task reviewer = new **`Agent`** with **Act as `code-reviewer`** (ZCode L2 review; not qc-specialist*), always via generic fallback `subagent_type: "general-purpose"` per C5 — no sticky resume unless host adds it later. Serial rule → **`parallel-dispatch.md`** § SDD implement.
+- **`Execution mode: sdd`**: one implementer **`Agent`** per task id (bare role id per C5, `general-purpose` fallback); task reviewer = new **`Agent`** with **Act as `code-reviewer`** (`subagent_type: "code-reviewer"`, `general-purpose` fallback; ZCode L2 review; not qc-specialist*), always with C5b prompt binding — no sticky resume unless host adds it later. Serial rule → **`parallel-dispatch.md`** § SDD implement.
 - **Never** multiple implementer Agents in one message for the same plan.
 
 ## Clarify
@@ -125,5 +153,5 @@ Cannot emit required **N** → **`Blocked`**.
 
 - Plugin install materializes source under `~/.zcode/cli/plugins/` — edit the harness checkout + reinstall to pick up harness changes.
 - Session todos (`TodoWrite`) are not durable SSOT unless mirrored to `{HARNESS_DIR}`.
-- No custom ZCode agent profiles for Morning Star roles — role binding is **always** prompt + skill load (C5b).
+- Morning Star roles **are** registered as ZCode subagent types from `agents/*.md` — but the shells are thin, so role binding is **still always** prompt + skill load (C5b); prefer bare role ids, the qualified form can fall back to the default prompt.
 - ZCode has no `sessionStart.skill`; new sessions do **not** auto-load PM — invoke `/morning-star-harness:pm` or `/skill:pm` manually.