Fish Aquatic Pet Health Diagnosis Analysis Tool | 鱼类水族宠物健康诊断分析工具

Fish aquatic pet health diagnosis analysis tool. When a user provides a video URL or file of aquatic pets such as goldfish, koi, betta, shrimp, crab, etc. fo...

Security Scan

The code and files mostly implement a fish/video analysis skill, but the SKILL.md and runtime behavior contain copy/paste errors, mismatched environment variable names, and storage/identity rules that do not align cleanly with the code — you should confirm a few things before installing.

Purpose & Capability

The repository contains scripts and API wrappers that match a video-based aquatic-pet health analysis (scripts/aquarium_analysis.py, ApiService, formatting functions). However the package also includes a separate face_analysis skill and large shared 'smyx_common' code, and many constant names (e.g., AUTISM_ANALYSIS, multiple scene codes) suggest copy/paste reuse from other domains. The inclusion of unrelated modules is explainable (shared library reuse) but increases the attack surface and is unexpected for a narrowly-scoped 'aquarium analysis' skill.

Instruction Scope

SKILL.md prescribes strict runtime rules (forbid reading local memory/LanceDB, mandatory open-id retrieval order, auto-saving attachments to an attachments directory) but contains clear inconsistencies: it tells to call 'python -m scripts.autism_analysis' (typo) while the script is scripts/aquarium_analysis.py; it requires retrieving OPENCLAW_SENDER_ID/sender_id but the code reads different environment names (OPENCLAW_SENDER_OPEN_ID / OPENCLAW_SENDER_USERNAME / FEISHU_OPEN_ID). The doc mandates auto-saving uploads to skill attachments, but I see no code that enforces saving uploaded attachments (the CLI accepts a local path or URL and the skill reads files), so the instruction set and implementation disagree.

Install Mechanism

No install spec is provided (instruction-only install), but the repo contains many Python modules and two large requirements.txt lists (face_analysis and smyx_common) enumerating many dependencies. The absence of an install step or declared dependencies in the SKILL metadata is inconsistent with the code footprint and could lead to missing dependency surprises or implicit installation of many packages if the operator attempts to run it.

Credentials

SKILL.md expects to acquire an 'open-id' from message context or env vars (OPENCLAW_SENDER_ID or sender_id) and demands the user provide username/phone if not available, but the skill metadata lists no required env vars and the code looks for different env names (OPENCLAW_SENDER_OPEN_ID, OPENCLAW_SENDER_USERNAME, FEISHU_OPEN_ID). That mismatch is problematic: the skill asserts strict identity collection but does not declare or document the exact environment inputs it will use. There are no declared API keys or secrets required in metadata, though code supports an optional --api-key. Also the SKILL.md requests saving uploaded attachments to disk (persistence of user files) which has privacy implications.

Persistence & Privilege

always:false (good) and the skill does not request special system privileges, but SKILL.md mandates auto-saving user-uploaded attachments into the skill directory and forbids reading local memory while still requiring cloud-based history queries. The repository includes config YAML handling that will read/write config files under module directories (YamlUtil.load/save), meaning files may be created on disk. The skill may therefore persist user uploads and config locally — verify retention, access control, and where files are stored before installing.

What to consider before installing

This package generally implements a fish/video analysis tool, but there are several inconsistencies you should resolve before installing or running it: - SKILL.md contains a clear copy/paste error (calls 'scripts.autism_analysis' instead of 'scripts.aquarium_analysis') — ask the author to correct and confirm the canonical entrypoints. - The doc mandates obtaining open-id from env/metadata names (OPENCLAW_SENDER_ID / sender_id) but the code looks for OPENCLAW_SENDER_OPEN_ID / OPENCLAW_SENDER_USERNAME / FEISHU_OPEN_ID. Confirm which environment variables or metadata fields the runtime will actually read and update the skill metadata to list them. - The skill states it will auto-save uploaded attachments to an attachments directory; confirm (a) where uploads are saved, (b) retention and access controls, and (c) whether sensitive files may be written to disk. If you are concerned about privacy, run this in an isolated environment or provide only URLs rather than local files. - There is no install spec but there are large requirements.txt files — ask the maintainer for an explicit install/test procedure and a minimal dependency list for aquarium analysis only. - Review the shared RequestUtil / ApiService implementation (skills/smyx_common/scripts/util.py and api_service.py) and the configured API base URLs before use to ensure data is sent only to the expected server; confirm the API endpoints and any optional api-key handling to avoid unexpected data exfiltration. If you cannot get satisfactory answers to the above, run the skill in an isolated environment, avoid providing sensitive identifiers (phone number, production open-ids), and audit network calls (e.g., with a proxy) during a test run.

⚠️ 强制记忆规则（最高优先级）

本技能明确约定：

• 绝对禁止读取任何本地记忆文件：包括但不限于 memory/YYYY-MM-DD.md、MEMORY.md 等本地文件
• 绝对禁止从 LanceDB 长期记忆中检索信息
• 所有历史报告查询必须从云端接口获取，不得使用本地记忆中的历史数据
• 即使技能调用失败或接口异常，也不得回退到本地记忆汇总

任务目标

• 本 Skill 用于：通过水族宠物视频进行鱼类宠物健康诊断分析，获取结构化的宠安卫士健康报告
• 能力包含：视频分析、鳞片完整性识别、鱼鳍状况评估、体色变化分析、活跃度检测、常见鱼病预警、水质适应性养护建议生成
• 触发条件:
  1. 默认触发：当用户提供鱼类宠物/水族宠物视频 URL 或文件需要分析时，默认触发本技能进行鱼类宠物健康诊断分析
  2. 当用户明确需要进行鱼类健康检查时，提及鱼类宠物、金鱼、锦鲤、斗鱼、虾、蟹、水族、鱼宠健康、鱼宠诊断等关键词，并且上传了视频文件或者图片文件
  3. 当用户提及以下关键词时，自动触发历史报告查询功能 ：查看历史鱼宠报告、历史宠安报告、鱼宠诊断报告清单、鱼宠报告清单、查询历史报告、查看鱼宠报告列表、显示所有鱼宠报告、显示鱼宠诊断报告，查询宠安卫士健康报告
• 自动行为：
  1. 如果用户上传了附件或者视频/图片文件，则自动保存到技能目录下 attachments
  2. ⚠️ 强制数据获取规则（次高优先级）：如果用户触发任何历史报告查询关键词（如"查看所有鱼宠报告"、"显示所有宠安报告"、" 查看历史报告"等），必须：
     • 直接使用 python -m scripts.autism_analysis --list --open-id {从消息上下文获取 open-id} 参数调用 API 查询云端的历史报告数据
     • 严格禁止：从本地 memory 目录读取历史会话信息、严格禁止手动汇总本地记录中的报告、严格禁止从长期记忆中提取报告
     • 必须统一从云端接口获取最新完整数据，然后以 Markdown 表格格式输出结果
     • 如果用户未明确提供 open-id，优先从 OpenClaw 消息上下文获取 sender id（如 metadata 中的 id 字段），然后尝试从当前消息上下文的环境变量 OPENCLAW_SENDER_ID 或者 sender_id 获取，无法获取时则必须用户提供用户名或者手机号作为 open-id

前置准备

• 依赖说明:scripts 脚本所需的依赖包及版本

  requests>=2.28.0
  

操作步骤

🔒 open-id 获取流程控制（强制执行，防止遗漏）

在执行鱼类宠物健康分析前，必须按以下优先级顺序获取 open-id：

第 1 步：检查用户是否在消息中明确提供了 open-id
        ↓ (未提供)
第 2 步：从当前消息上下文的环境变量中获取 OPENCLAW_SENDER_ID
        ↓ (无法获取)
第 3 步：从当前消息上下文的环境变量中获取 sender_id
        ↓ (无法获取)
第 4 步：从 OpenClaw 消息元数据中获取 id 字段（如 metadata 中的 id/session_id/user_id等）作为 open-id
        ↓ (无法获取)
第 5 步：❗ 必须暂停执行，明确提示用户提供用户名或手机号作为 open-id

⚠️ 关键约束：

• 禁止自行假设或生成 open-id 值（如 fishC113、fish123 等）
• 禁止跳过 open-id 验证直接调用 API
• 必须在获取到有效 open-id 后才能继续执行分析
• 如果用户拒绝提供 open-id，说明用途（用于保存和查询鱼宠报告记录），并询问是否继续

---

• 标准流程:
  1. 准备视频输入
     • 提供本地视频文件路径或网络视频 URL
     • 确保视频清晰展示鱼儿整体外观、鳞片、鱼鳍、游动姿态，光线充足
  2. 获取 open-id（强制执行）
     • 按上述流程控制获取 open-id
     • 如无法获取，必须提示用户提供用户名或手机号
  3. 执行鱼类宠物健康分析
     • 调用 -m scripts.aquarium_analysis 处理视频文件（必须在技能根目录下运行脚本）
     • 参数说明:
       • --input: 本地视频文件路径（使用 multipart/form-data 方式上传）
       • --url: 网络视频 URL 地址（API 服务自动下载）
       • --fish-type: 鱼类宠物类型，可选值：goldfish/koi/betta/shrimp/crab/turtle/clownfish/guppy/arowana/angel/other，默认 other
       • --open-id: 当前用户的 OpenID/UserId（必填，按上述流程获取）
       • --list: 显示鱼类宠物视频历史分析报告列表清单（可以输入起始日期参数过滤数据范围）
       • --api-key: API 访问密钥（可选）
       • --api-url: API 服务地址（可选，使用默认值）
       • --detail: 输出详细程度（basic/standard/json，默认 json）
       • --output: 结果输出文件路径（可选）
  4. 查看分析结果
     • 接收结构化的宠安卫士健康报告
     • 包含：鱼类宠物基本信息、整体健康状况、鳞片分析、鱼鳍状态、体色分析、潜在疾病预警、健康养护建议

资源索引

• 必要脚本：见 scripts/aquarium_analysis.py(用途：调用 API 进行水族宠物健康分析，本地文件使用 multipart/form-data 方式上传，网络 URL 由 API 服务自动下载)
• 配置文件：见 scripts/config.py(用途：配置 API 地址、默认参数和视频格式限制)
• 领域参考：见 references/api_doc.md(何时读取：需要了解 API 接口详细规范和错误码时)

注意事项

• 仅在需要时读取参考文档，保持上下文简洁
• 视频要求：支持 mp4/avi/mov 格式，最大 100MB
• API 密钥可选，如果通过参数传入则必须确保调用鉴权成功，否则忽略鉴权
• 分析结果仅供健康参考，不能替代专业宠医诊断
• 禁止临时生成脚本，只能用技能本身的脚本
• 传入的网路地址参数，不需要下载本地，默认地址都是公网地址，api 服务会自动下载
• 当显示历史分析报告清单的时候，从数据 json 中提取字段 reportImageUrl 作为超链接地址，使用 Markdown 表格格式输出，包含" 报告名称"、"鱼宠类型"、"分析时间"、"点击查看"四列，其中"报告名称"列使用鱼宠健康分析报告-{记录id}形式拼接, "点击查看"列使用 [🔗 查看报告](reportImageUrl) 格式的超链接，用户点击即可直接跳转到对应的完整报告页面。
• 表格输出示例：

  报告名称	鱼宠类型	分析时间	点击查看
  鱼宠健康分析报告 -20260312172200001	金鱼	2026-03-12 17:22:00	🔗 查看报告

使用示例

# 分析本地金鱼视频（OpenClaw UI 上下文，使用 metadata id 作为 open-id）
python -m scripts.aquarium_analysis --input /path/to/goldfish_video.mp4 --fish-type goldfish --open-id openclaw-control-ui

# 分析网络锦鲤视频（OpenClaw UI 上下文，使用 metadata id 作为 open-id）
python -m scripts.aquarium_analysis --url https://example.com/koi_video.mp4 --fish-type koi --open-id openclaw-control-ui

# 分析本地斗鱼视频（OpenClaw UI 上下文，使用 metadata id 作为 open-id）
python -m scripts.aquarium_analysis --input /path/to/betta_video.mp4 --fish-type betta --open-id openclaw-control-ui

# 分析本地观赏虾视频（OpenClaw UI 上下文，使用 metadata id 作为 open-id）
python -m scripts.aquarium_analysis --input /path/to/shrimp_video.mp4 --fish-type shrimp --open-id openclaw-control-ui

# 分析本地螃蟹视频（OpenClaw UI 上下文，使用 metadata id 作为 open-id）
python -m scripts.aquarium_analysis --input /path/to/crab_video.mp4 --fish-type crab --open-id openclaw-control-ui

# 分析本地乌龟视频（OpenClaw UI 上下文，使用 metadata id 作为 open-id）
python -m scripts.aquarium_analysis --input /path/to/turtle_video.mp4 --fish-type turtle --open-id openclaw-control-ui

# 显示历史分析报告/显示分析报告清单列表/显示历史宠安报告（自动触发关键词：查看历史鱼宠报告、历史报告、鱼宠报告清单等）
python -m scripts.aquarium_analysis --list --open-id openclaw-control-ui

# 输出精简报告
python -m scripts.aquarium_analysis --input video.mp4 --fish-type goldfish --open-id your-open-id --detail basic

# 保存结果到文件
python -m scripts.aquarium_analysis --input video.mp4 --fish-type koi --open-id your-open-id --output result.json