What makes Aicwos notable?

990 downloads. Trust evidence available

How should Aicwos be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

Aicwos

MANDATORY data pipeline for 口播文案/短视频脚本/系列口播/讲师风格学习/知识库管理 — LLM CANNOT access lecturer profiles, product knowledge, or series progress without this skill. Act... Skill: Aicwos Owner: smallroya Summary: MANDATORY data pipeline for 口播文案/短视频脚本/系列口播/讲师风格学习/知识库管理 — LLM CANNOT access lecturer profiles, product knowledge, or series progress without this skill. Act... Tags: latest:1.10.5 Version history: v1.10.5 | 2026-04-16T10:06:59.065Z | user fix: 清理遗留问题 — workflow-lecturer.md 对齐 staging 路径和 --save-sample; 移除 heredoc 语法; 移除 manifest.json 残留引用; 移除签名认证代码 v1.10.4 | 2026-04-16T10:00:4

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

990

Updated

Oct 11, 2026

Version

1.10.5

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 990 downloads reported by the source. Last updated 10/11/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 11, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 11, 2026
Adoption signal: 990 downloadsadoption · observed Oct 11, 2026
Latest release: 1.10.5release · observed Apr 16, 2026
Handshake status: UNKNOWNsecurity

What changed

Newest first. Only changes that point at a source are listed.

Release 1.10.5release
fix: 清理遗留问题 — workflow-lecturer.md 对齐 staging 路径和 --save-sample; 移除 heredoc 语法; 移除 manifest.json 残留引用; 移除签名认证代码
Observed Apr 16, 2026 · source

Install and run

Setup complexity: low.

clawhub skill install s17aey8rv6b658ncwxxnc44feh84hm11:aicwos

Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.
Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-smallroya-aicwos/snapshot"

Run-check

$0.02 USD

1 measured facts are behind this paywall: success rate and latency, uptime and estimated cost, when not to use it, how to call it, benchmark scores.

Agents pay $0.02 in USDC. A card payment is $0.50, the smallest a card allows.

Documentation

CLAWHUB

158,934 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: aicwos
version: 1.10.5
description: >
  MANDATORY data pipeline for 口播文案/短视频脚本/系列口播/讲师风格学习/知识库管理 —
  LLM CANNOT access lecturer profiles, product knowledge, or series progress without
  this skill. Activate when user says "帮我写口播", "写文案", "写脚本", "写个短视频",
  "学一下讲师风格", "继续写系列", "生成系列口播", "学习讲师", "知识库", "看看讲师",
  "系列进度", "改一下第X集", or ANY request involving copywriting, lecturer, series, or knowledge base.
  Provides precise style replication, knowledge retrieval, series continuity, and
  persistent storage that LLM intrinsically lacks.
dependency:
  python:
    - jieba>=0.42.1
    - onnxruntime>=1.16.0
    - numpy>=1.24.0
---

# Aicwos - Mandatory Copywriting Data Pipeline

## MANDATORY RULES (NON-NEGOTIABLE)

1. **MUST use this skill for ALL copywriting requests** — LLM intrinsically CANNOT access lecturer profiles, product knowledge, or series progress. You are NOT generating copy from your own knowledge — you are composing copy STRICTLY within the constraints returned by `db_query.py --type context`. NEVER write copy without calling `--type context` first.

2. **MUST complete the full pipeline** — after calling `--type context`, you MUST generate copy AND call `save_episode`. NEVER retrieve context then skip saving. Partial usage leaves orphan data and breaks series continuity.

3. **MUST use `db_query.py` and `lecturer_analyzer.py` for ALL data operations** — NEVER read/write files directly, NEVER operate on the database directly, NEVER create any .py/.json files on the user's computer. If the CLI needs long JSON, use `--query-file` instead of `--query`.

4. **Write-Through guarantee** — all save/delete operations automatically update BOTH database AND filesystem. You do NOT need to call `db_sync.py` after save/delete operations. The sync tool is only for initial bulk import and recovery.

## CORE CONCEPTS

| Term | Definition |
|------|-----------|
| 轻量画像 (lite profile) | Abbreviated lecturer profile (~400 tokens) with style dimensions and persona mapping. Retrieved via `--action lite`. Used during copy generation — never load full profile. |
| 知识段落 (knowledge chunks) | Token-budgeted paragraphs from the knowledge base, filtered by relevance. Retrieved via `--type context` or `knowledge --action context`. |
| 摘要衔接 (summary continuity) | Each episode stores a `summary` (~100 chars) and `hook_next`. When generating episode N, read summaries of episodes 1..N-1 via `--action summaries` — NEVER read full episode text. |
| 行为规则 (behavior rules) | Forbidden patterns (B-prefixed) and required patterns (R-prefixed). Loaded automatically by `--type context`. |
| 系列进度 (series progress) | `completed/total_episodes` tracked in DB. Updated automatically by `save_episode` and `delete_episode`. Do NOT compute manually. |
| Write-Through (双写) | Every save/delete operation updates both DB and filesystem atomically. No manual sync needed. DB is the query layer; filesystem is the human-readable layer. |

## TRIGGER CONDITIONS

Activate this skill when user

_meta.json

{
  "ownerId": "kn7f31nc7qnrvrsja0yz0kpay584hxma",
  "slug": "aicwos",
  "version": "1.10.5",
  "publishedAt": 1776334019065
}

references/profile-format.md

# 画像格式规范

画像数据存储在SQLite，按需加载字段。以下定义profile_json完整结构。

## 核心字段

| 字段 | 类型 | 说明 | 加载时机 |
|------|------|------|---------|
| profile_version | string | "2.0" | 始终 |
| lecturer_id | string | 讲师标识 | 始终 |
| lecturer_name | string | 显示名称 | 始终 |
| qualitative | object | 定性分析 | 文案生成 |
| persona_mapping | object | 人设映射 | 文案生成 |
| style_dimensions | object | 文风五维 | 文案生成 |
| reference_scripts | array | 参考示例索引 | 文案生成 |
| sample_excerpts | object | 精彩片段 | 文案生成 |
| quantitative | object | 定量分析 | 增量合并 |
| sample_stats | object | 样本统计 | 管理 |
| merge_history | array | 合并历史 | 管理 |
| created_at | string | 创建时间 | 管理 |
| updated_at | string | 更新时间 | 管理 |

## Token优化对比

| 场景 | 全量加载 | 按需加载 | 节省 |
|------|----|----|------|
| 文案生成（加载画像） | ~2500 tokens | ~400 tokens | 84% |
| 增量合并（加载定量） | ~2500 tokens | ~1500 tokens(单独) | 40% |
| 知识检索(1产品5文档) | ~1500 tokens | ~300-500 tokens | 67-80% |
| 系列连续性(10集) | ~3200 tokens | ~600 tokens(摘要) | 81% |

## quantitative字段合并策略

| 字段 | 策略 | 说明 |
|------|------|------|
| basic_stats | 计数型累加 | total_chinese_chars等直接相加 |
| sentence_length | 数值型加权平均 | avg/median/stdev按权重合并 |
| word_frequency | 列表型合并去重 | 同词累加count，取Top50 |
| bigrams | 列表型合并去重 | 同上 |
| punctuation | 计数型累加+比率加权 | counts累加，ratios加权 |
| vocabulary_diversity | 数值型加权平均 | ttr按权重合并 |
| rhetorical_patterns | 计数型累加+比率加权 | 同punctuation |
| connector_patterns | 计数型累加 | 各类别count累加 |
| opening/closing_patterns | 列表保留最新5条 | 新数据覆盖旧数据 |

references/script-styles.md

# 文案风格参考

## 风格类型

| 风格 | 特点 | 适合讲师 |
|------|------|---------|
| 干货型 | 信息密度高、逻辑清晰、观点明确 | 正式度4-5、情感1-2、节奏3-4 |
| 故事型 | 情节驱动、代入感强、有转折 | 正式度2-3、情感3-4、节奏2-3 |
| 对话型 | 亲切互动、提问引导、像聊天 | 正式度1-2、情感3-4、互动4-5 |
| 激励型 | 情绪饱满、有号召力、节奏快 | 正式度2-3、情感4-5、节奏4-5 |
| 知识型 | 深度解析、引用数据、逻辑链长 | 正式度4-5、深度4-5、节奏2-3 |

## 文风五维

每个维度1-5级，1=最低，5=最高：

| 维度 | 1级 | 3级 | 5级 |
|------|-----|-----|-----|
| formality(正式度) | 口语/网络用语 | 日常对话 | 书面/专业术语 |
| emotion(情感度) | 冷静客观 | 适度情感 | 强烈情绪/共情 |
| pace(节奏) | 慢/每句15字+ | 中等/混合 | 快/短句为主 |
| interaction(互动) | 无互动 | 偶尔提问 | 频繁提问/号召 |
| depth(深度) | 表面泛谈 | 有论据支撑 | 深度分析/数据 |

## 字数结构

| 位置 | 占比 | 内容 |
|------|------|------|
| 开场Hook | 10-15% | 抓注意力、设悬念 |
| 主体内容 | 60-70% | 核心信息、论据、案例 |
| 结尾CTA | 10-20% | 总结、号召、预告 |

目标字数偏差不超过10%。口播600字约2分钟。

references/workflow-copywriting.md

# 文案生成流程

架构核心设计：
1. 文案生成时只加载轻量画像（~400 tokens），不全量读取profile.json
2. 知识库精准检索，只返回相关段落，不读整个文件
3. 系列文案连续性只读摘要，不读全文
4. 所有写入操作通过 `db_query.py` CLI 完成，智能体不可自行写文件或创建脚本

## 获取生成上下文

一条命令获取文案生成所需的全部数据：

```bash
python scripts/db_query.py --type context --lecturer 讲师A --query "养生保健" --data-dir <控制台目录>
```

返回：
- `profile`: 轻量画像（lecturer_name + qualitative + persona_mapping + style_dimensions）
- `reference_samples`: 参考示例
- `knowledge_context`: token预算内的精准知识段落
- `behavior_rules`: 行为规则（允许+禁止）

## 单篇文案生成

1. 获取生成上下文（上述命令）
2. 加载风格模板：见 [script-styles.md](script-styles.md)
3. 融合画像+参考示例+知识段落+风格 → 生成文案
4. 保存文案（单篇视为1集系列）：
   ```bash
   # 先创建系列计划（1集）
   python scripts/db_query.py --type series --action save_plan \
     --id "<文案主题>" --lecturer <讲师> \
     --query '{"title":"<主题>","episodes":[{"num":1,"title":"<标题>","topic":"<主题>"}]}' \
     --data-dir <控制台目录>
   # 再保存文案
   python scripts/db_query.py --type series --action save_episode \
     --id "<文案主题>" --id2 1 \
     --query '{"title":"<标题>","content":"<正文>","summary":"<100字摘要>","hook_next":""}' \
     --data-dir <控制台目录>
   ```

文案保存到 `控制台/讲师列表/{讲师}/系列文案/{主题}/E01_{标题}.txt`，路径由脚本强制控制。

## 系列文案

1. 获取生成上下文
2. 制定系列计划（智能体生成每集标题和主题）
3. 保存系列计划到数据库：
   ```bash
   python scripts/db_query.py --type series --action save_plan \
     --id <系列名> --lecturer <讲师> \
     --query '{"title":"<系列标题>","episodes":[{"num":1,"title":"第1集标题","topic":"主题"},{"num":2,"title":"第2集标题","topic":"主题"}]}' \
     --data-dir <控制台目录>
   ```
4. 查看系列计划：
   ```bash
   python scripts/db_query.py --type series --action get_plan --id <系列名> --data-dir <控制台目录>
   ```
5. 逐集生成，每集保存：
   ```bash
   python scripts/db_query.py --type series --action save_episode \
     --id <系列名> --id2 <集号> \
     --query '{"title":"<标题>","content":"<正文>","summary":"<100字摘要>","hook_next":"<下集衔接点>"}' \
     --data-dir <控制台目录>
   ```

**连续性优化**：生成第N集时只读前N-1集的摘要（每集~100字），不读全文。

```bash
# 查看系列摘要（连续性用）
python scripts/db_query.py --type series --action summaries --id <系列名> --data-dir <控制台目录>
```

## 修订流程（改集/删集）

当用户要求修改某集文案时，必须先删除旧数据再保存新数据。

### 修改第X集

1. 重新获取生成上下文（不要复用旧context）：
   ```bash
   python scripts/db_query.py --type context --lecturer <讲师> --query "<主题>" --data-dir <控制台目录>
   ```
2. 读取当前文案（了解原文内容）：
   ```bash
   python scripts/db_query.py --type series --action content --id <系列名> --id2 X --data-dir <控制台目录>
   ```
3. 删除旧集数据（必须！否则DB出现重复记录）：
   ```bash
   python scripts/db_query.py --type series --action delete_episode --id <系列名> --id2 X --data-dir <控制台目录>
   ```
4. 根据用户要求修改文案
5. 保存修改后的文案：
   ```bash
   python scripts/db_query.py --type series --action save_episode \
     --id <系列名> --id2 X \
     --query '{"title":"<标题>","content":"<正文>","summary":"<100字摘要>","hook_next":"<下集衔接>"}' \
     --data-dir <控制台目录>
   ```

**关键**：`delete_episode` 必须在 `save_episode` 之前执行。跳过 `delete_episode` 会导致DB重复记录和旧的.txt文件残留。

进度计数器自动处理：修订中间集（如第2集）不会导致进度倒退，因为进度取的是 MAX(episode_num)。

### 删除某集

```bash
python scripts/db_query.py --type series --action delete_episode --id <系列名> --id2 X --data-dir <控制台目录>

Github ReposUpdated 3d agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.

MCPOPENCLAW

Github ReposUpdated 7mo agoRank 70

CopilotKit

The Frontend for Agents & Generative UI. React + Angular