Ecosystem Skill — Supports building and managing the MOVA ecosystem. Requires the openclaw-mova plugin.
MOVA Spec Guide
Answer any question about the MOVA language — from basic concepts to schema-level validation — by reading the canonical spec documents in the workspace and referencing the public repository.
What this skill does
- 1. Explains MOVA concepts — schemas (
ds.*), envelopes (env.*), verbs, episodes, global catalogs, instruction profiles, runtime bindings, connectors, security layer, text/UI layer - Reads authoritative source docs — uses the local spec clone for immediate access; references the public GitHub repo for the canonical latest state
- Shows examples — points to
examples/ files and explains them in context - Validates artifacts — given a JSON draft, checks it against the relevant MOVA schema and explains violations
- Maps concepts to practice — explains how a concept applies to a real use case the user brings
Source hierarchy
| Priority | Source | Use for |
|---|
| 1 (primary) | Local workspace: INLINECODE4 | All reads — fast and always available |
| 2 (canonical reference) |
GitHub:
https://github.com/mova-compact/mova-spec | Latest version check, stable URLs to share |
Always read from the local workspace first. Only reference the GitHub repo if the user needs a link to share or asks about the latest uncommitted changes.
Workspace layout
CODEBLOCK0
Core MOVA concepts (quick map)
| Concept | What it is | Where defined |
|---|
| INLINECODE6 schema | JSON Schema describing what data looks like | INLINECODE7 |
| INLINECODE8 envelope |
Typed speech-act: a request, command, or event over
ds.* data |
schemas/env.*.schema.json |
| Verb | Abstract operation type: create, update, route, record, publish, analyze, ... |
docs/mova_global_and_verbs.md |
| Tool | Execution channel (
tool_id);
0 = tool-less |
docs/mova_global_and_verbs.md §4.5 |
| Action signature | Atomic unit for policy:
(verb_id, tool_id, target_kind?) |
MOVA_6.0.0_RELEASE_NOTES.md |
| Episode | Structured record of one meaningful work step |
ds.mova_episode_core_v1 |
| Instruction profile | Declarative policy and guardrail set for an executor |
ds.instruction_profile_core_v1 |
| Global catalog | Shared vocabulary for all layers |
global.*.json |
| Operator frame | 13-axis audit lens (what/how/where/when/why/...) |
docs/mova_operator_frame.md |
| Genetic layer | Pattern memory built from episodes over time |
docs/mova_episodes_and_genetic_layer.md |
How to handle different question types
Concept question
Read the relevant
docs/ file. Summarize the answer. Quote the exact definition. Show a minimal example if one exists in
examples/.
Schema question
Read the relevant
schemas/*.schema.json. Explain each field: type, required/optional, allowed values. Show how the field relates to the concept.
Validation question
Read the user's JSON. Read the target schema. Check required fields, field types, allowed enum values. Report each violation with the relevant schema field and rule.
How to write question
Read the most relevant
docs/ file. Walk through a minimal valid instance step by step. Reference the
examples/ file if it exists.
Catalog lookup
Read the relevant
global.*.json. Quote the exact entry. Explain the purpose in one sentence.
MOVA version
Current version in the workspace: MOVA 6.0.0
Key additions in 6.0.0 (see MOVA_6.0.0_RELEASE_NOTES.md):
- - Normative definition of
verb, tool, INLINECODE31 - INLINECODE32 is canonical for tool-less actions
- Policy matching priority: INLINECODE33
- INLINECODE34 in domain dictionaries (readability, no normative weight)
Rules
- - NEVER invent schema field names, verb IDs, or catalog values — always read from workspace files
- NEVER reference schema URLs without first reading the local file to confirm the content
- If the user asks about a concept not yet defined in the spec — say so explicitly, do not improvise
- Always cite which document and section the answer comes from
- If the user provides a JSON artifact to validate — read the correct schema file first, then compare field by field
- MOVA does not execute — never describe MOVA artifacts as having runtime behavior
生态系统技能 — 支持构建和管理 MOVA 生态系统。需要 openclaw-mova 插件。
MOVA 规范指南
通过阅读工作区中的权威规范文档并引用公共仓库，回答关于 MOVA 语言的任何问题——从基本概念到模式级验证。
本技能的功能
- 1. 解释 MOVA 概念 — 模式 (ds.)、信封 (env.)、动词、事件片段、全局目录、指令配置文件、运行时绑定、连接器、安全层、文本/UI 层
- 读取权威源文档 — 使用本地规范克隆进行即时访问；引用公共 GitHub 仓库获取规范的最新状态
- 展示示例 — 指向 examples/ 文件并在上下文中进行解释
- 验证工件 — 给定 JSON 草稿，对照相关 MOVA 模式进行检查并解释违规情况
- 将概念映射到实践 — 解释概念如何应用于用户带来的实际用例
源层次结构
| 优先级 | 源 | 用途 |
|---|
| 1（主要） | 本地工作区：/home/mova/.openclaw/workspace/mova-spec/ | 所有读取 — 快速且始终可用 |
| 2（规范参考） |
GitHub：https://github.com/mova-compact/mova-spec | 最新版本检查，可共享的稳定 URL |
始终优先从本地工作区读取。 仅当用户需要可共享的链接或询问最新的未提交更改时，才引用 GitHub 仓库。
工作区布局
mova-spec/
docs/
mova_core.md ← 核心规范（概念问题从此处开始）
movaglobalandverbs.md ← 动词、全局目录、actionsignature
movaoperatorframe.md ← 操作者框架：13 轴审计视角
movaepisodesandgeneticlayer.md← 事件片段结构和模式记忆
movalayersand_namespaces.md ← 红色核心/技能/基础设施分层
movaruntimeand_connectors.md ← 运行时绑定和连接器契约
movasecuritylayer.md ← 指令配置文件和安全事件
movatextanduilayer.md ← 文本通道分离
MOVA6.0.0RELEASENOTES.md ← 最新更改：动词/工具/actionsignature
schemas/
ds.movaschemacore_v1.schema.json
ds.movaepisodecore_v1.schema.json
ds.instructionprofilecore_v1.schema.json
ds.connectorcorev1.schema.json
ds.runtimebindingcore_v1.schema.json
ds.uitextbundlecorev1.schema.json
ds.mova4corecatalog_v1.schema.json
ds.securityeventepisodecorev1.schema.json
env.*.schema.json（信封）
examples/
mova4corecatalog.example.json
action_signature.example.json
env.*.example.json
global.episodetypecatalog_v1.json
global.securitycatalogv1.json
global.layersandnamespaces_v1.json
global.textchannelcatalog_v1.json
README-mova-spec.md
核心 MOVA 概念（快速映射）
| 概念 | 定义 | 定义位置 |
|---|
| ds. 模式 | 描述数据样子的 JSON Schema | schemas/ds..schema.json |
| env. 信封 |
类型化言语行为：针对 ds. 数据的请求、命令或事件 | schemas/env.*.schema.json |
| 动词 | 抽象操作类型：创建、更新、路由、记录、发布、分析…… | docs/mova
globaland_verbs.md |
| 工具 | 执行通道 (tool
id)；0 = 无工具 | docs/movaglobal
andverbs.md §4.5 |
| Action signature | 策略的原子单元：(verb
id, toolid, target
kind?) | MOVA6.0.0
RELEASENOTES.md |
| 事件片段 | 一个有意义工作步骤的结构化记录 | ds.mova
episodecore_v1 |
| 指令配置文件 | 执行者的声明性策略和护栏集 | ds.instruction
profilecore_v1 |
| 全局目录 | 所有层的共享词汇表 | global.*.json |
| 操作者框架 | 13 轴审计视角（什么/如何/哪里/何时/为什么/……） | docs/mova
operatorframe.md |
| 遗传层 | 随时间从事件片段构建的模式记忆 | docs/mova
episodesand
geneticlayer.md |
如何处理不同类型的问题
概念问题
读取相关的 docs/ 文件。总结答案。引用确切定义。如果 examples/ 中存在，则展示一个最小示例。
模式问题
读取相关的 schemas/*.schema.json。解释每个字段：类型、必需/可选、允许的值。展示该字段如何与概念关联。
验证问题
读取用户的 JSON。读取目标模式。检查必需字段、字段类型、允许的枚举值。报告每个违规情况及其相关的模式字段和规则。
如何编写问题
读取最相关的 docs/ 文件。逐步构建一个最小有效实例。如果存在，则引用 examples/ 文件。
目录查找
读取相关的 global.*.json。引用确切的条目。用一句话解释其用途。
MOVA 版本
工作区中的当前版本：MOVA 6.0.0
6.0.0 中的关键新增内容（参见 MOVA6.0.0RELEASE_NOTES.md）：
- - verb、tool、actionsignature 的规范性定义
- toolid = 0 是无工具操作的规范值
- 策略匹配优先级：actionsignature > verbid > toolid
- 领域字典中的 actionlabels（可读性，无规范性权重）
规则
- - 绝不发明模式字段名称、动词 ID 或目录值 — 始终从工作区文件读取
- 绝不引用模式 URL 而不先读取本地文件以确认内容
- 如果用户询问规范中尚未定义的概念 — 明确说明，不要即兴发挥
- 始终引用答案来自哪个文档和章节
- 如果用户提供要验证的 JSON 工件 — 先读取正确的模式文件，然后逐字段比较
- MOVA 不执行 — 绝不描述 MOVA 工件具有运行时行为
