← Back to Skills Marketplace
Fixed Amount Invoice Ocr
by
SCNet-sugon
· GitHub ↗
· v1.0.4
· MIT-0
116
Downloads
0
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install fixed-amount-invoice-ocr
Description
支持定额发票识别。
README (SKILL.md)
\r
Sugon-Scnet 定额发票识别 OCR 技能\r
\r 本技能封装了 Sugon-Scnet 定额发票识别 OCR 服务,通过单一接口即可调用 1 种识别能力,高效提取文字及票据信息。\r \r
功能特性\r
\r
- 定额票据:覆盖定额发票识别,自动提取关键字段。\r \r
前置配置\r
\r
⚠️ 重要:使用前需要申请 Scnet API Token\r \r
申请 API Token\r
\r
- 访问 Scnet 官网 注册/登录\r
- 在控制台申请 API 密钥(格式:
sc-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx)\r - 复制密钥备用\r \r
配置 Token\r
\r 手动配置(推荐)\r
- 在技能目录下创建
config/.env文件,内容如下:\r
# ===== Sugon-Scnet OCR API 配置 =====\r
# 申请地址:https://www.scnet.cn\r
SCNET_API_KEY=your_scnet_api_key_here\r
\r
# API 基础地址(一般无需修改)\r
SCNET_API_BASE=https://api.scnet.cn/api/llm/v1\r
```\r
2. 添加:`SCNET_API_KEY=你的密钥`\r
3. 设置文件权限为 600(仅所有者可读写)\r
**⚠️ 安全警告**:切勿将 API Key 直接粘贴到聊天对话中,否则可能被记录或泄露。\r
\r
### Token 更新\r
\r
Token 过期后调用会返回 401 或 403 错误。更新方法:重新申请 Token 并替换 config/.env 中的 SCNET_API_KEY。\r
\r
### 依赖安装\r
\r
本技能需要 Python 3.6+ 和 requests 库。请运行以下命令:\r
\r
```bash\r
pip install requests\r
```\r
---\r
### 使用方法\r
\r
### 参数说明\r
\r
| 参数名 | 类型 | 必填 | 描述 |\r
|--------|------|------|------|\r
| ocrType | string | 是 | 识别类型枚举。必须为以下之一:\x3Cbr>• QUOTA_INVOICE(定额发票) |\r
| filePath | string | 是 | 待识别图片的本地绝对路径。支持 jpg、png、pdf 等常见格式。 |\r
\r
### 命令行调用示例\r
\r
```bash\r
python .claude/skills/fixed_amount_invoice_ocr/scripts/main.py QUOTA_INVOICE /path/to/invoice.jpg\r
```\r
\r
### 在 AI 对话中使用\r
\r
用户可以说:\r
\r
- “帮我识别这张定额发票,图片在 /Users/name/Downloads/invoice.jpg”\r
\r
AI 会根据 description 中的关键词自动触发本技能。\r
\r
### AI 调用建议\r
为避免触发 API 速率限制(10 QPS),请串行调用本技能,即等待前一个识别完成后再发起下一个请求。\r
如果使用 OpenClaw 的 exec 工具,建议设置 timeout 或 yieldMs 参数,让命令同步执行,避免多个命令同时运行导致并发。\r
\r
### 配置选项\r
\r
编辑 `config/.env` 文件:\r
\r
| 变量名 | 默认值 | 说明 |\r
|--------|--------|------|\r
| SCNET_API_KEY | 必需 | Scnet API 密钥 |\r
| SCNET_API_BASE | https://api.scnet.cn/api/llm/v1 | API 基础地址(一般无需修改) |\r
\r
### 输出\r
\r
- 标准输出:识别结果的 JSON 数据,结构与 API 文档一致,位于 `data` 字段内。\r
- 识别结果位于 data[0].result[0].elements 中,具体字段取决于 ocrType。\r
- 错误信息:如果发生错误,会输出以 `错误:` 开头的友好提示。\r
\r
### 注意事项\r
\r
- 本技能调用的 OCR API 有 10 QPS 的速率限制。\r
- 如果遇到 429 错误,请等待 2-3 秒后重试,不要连续发起请求。\r
- 建议在调用前确保图片已准备就绪,避免因网络问题导致重复调用。\r
\r
### 故障排除\r
\r
| 问题 | 解决方案 |\r
|------|----------|\r
| 配置文件不存在 | 创建 config/.env 并填入 Token(参考前置配置) |\r
| API Key 无效/过期 | 重新申请 Token 并更新 `.env` 文件 |\r
| 文件不存在 | 检查提供的文件路径是否正确 |\r
| 网络连接失败 | 检查网络连接或防火墙设置 |\r
| 不支持的文件类型 | 确保文件扩展名为允许的类型(参考 API 文档) |\r
| 401/403/Unauthorized | Token 无效或过期,重新申请并配置 |\r
| 429 Too Many Requests | 请求过于频繁,技能会自动等待并重试(最多 3 次)。若持续失败,请降低调用频率或联系服务方提高限额。 |\r
\r
Usage Guidance
This skill is internally consistent with its stated purpose: it reads a local image and uploads it to api.scnet.cn using a SCNET_API_KEY. Before installing, verify the source (the skill lists a GitHub homepage placeholder in skill.yaml—confirm the real repository), and fix the metadata discrepancy (registry saying no env vars vs. files declaring SCNET_API_KEY). Only provide a scoped API key you trust; do not paste the key into chat. Consider running the script in an isolated environment or container, restrict the API key permissions if possible, and review the small Python script yourself (it is included) to confirm there are no changes you object to. If you want higher assurance, ask the publisher to correct the registry metadata and provide a verifiable upstream repository URL.
Capability Tags
Capability Assessment
Purpose & Capability
Declared purpose is fixed‑amount invoice OCR and the code/skilled instructions call an OCR endpoint (https://api.scnet.cn/api/llm/v1/ocr/recognize) using an API key; this is coherent. However the registry metadata at the top of the report claimed no required env vars while SKILL.md and skill.yaml correctly declare SCNET_API_KEY as required—this mismatch is a packaging/documentation inconsistency.
Instruction Scope
SKILL.md and scripts/main.py limit actions to reading a local image file, loading SCNET_API_KEY (from env or config/.env), and making a multipart POST to the official Scnet OCR endpoint. The script prints JSON results and exits on errors. There are no instructions to read unrelated files, transmit data to third‑party endpoints beyond api.scnet.cn, or perform other system probing.
Install Mechanism
This is an instruction-only skill with a small Python script and no install spec; it only requires Python and the requests library (pip install requests). No remote archives or arbitrary downloads are executed by the skill itself.
Credentials
The skill requires a single service credential (SCNET_API_KEY) and optionally SCNET_API_BASE, which is appropriate for an HTTP API client. The concerning bit is the top-level registry metadata in the report that listed no required env vars while both SKILL.md and skill.yaml require SCNET_API_KEY—this mismatch should be resolved before trusting metadata-driven installs.
Persistence & Privilege
always:false and the skill does not request persistent/system-wide changes or alter other skills. It does not enable any elevated or always‑loaded behavior.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install fixed-amount-invoice-ocr - After installation, invoke the skill by name or use
/fixed-amount-invoice-ocr - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.4
fixed_amount_invoice_ocr 1.0.4
- 示例对话更新,更准确地展示定额发票识别的推荐用法。
- 旧版技能说明中的身份证、文字通用识别等表述已删除,使文档描述与技能功能一致。
- 其余内容未变。
v1.0.3
- Added skill.yaml file for clearer metadata and dependency management.
- Updated SKILL.md with required and optional environment variables, explicit Python and requests dependencies, and strong API Key safety warnings.
- Improved documentation on environment setup, security best practices, and dependency installation.
v1.0.1
Version 1.0.1
- No file changes detected; SKILL.md remains unchanged.
- No updates, bug fixes, or new features in this release.
v0.1.0
- Initial release of fixed-amount-invoice-ocr skill.
- Supports detection and structured extraction of key fields from fixed-amount invoices using Sugon-Scnet OCR.
- Requires SCNet API Token configuration.
- Accepts image file path and OCR type as input; outputs structured JSON with results and confidence scores.
- Includes usage instructions, configuration steps, rate limit guidance, and troubleshooting tips.
Metadata
Frequently Asked Questions
What is Fixed Amount Invoice Ocr?
支持定额发票识别。 It is an AI Agent Skill for Claude Code / OpenClaw, with 116 downloads so far.
How do I install Fixed Amount Invoice Ocr?
Run "/install fixed-amount-invoice-ocr" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Fixed Amount Invoice Ocr free?
Yes, Fixed Amount Invoice Ocr is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Fixed Amount Invoice Ocr support?
Fixed Amount Invoice Ocr is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Fixed Amount Invoice Ocr?
It is built and maintained by SCNet-sugon (@scnet-sugon); the current version is v1.0.4.
More Skills