Prana A股财务分析助手

v1.0.4

通过调用Prana平台上的远程 agent 执行 A股财务分析助手,并将结果返回给调用方

0· 129·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for gocybertrade/prana-stock-scoring-analysis.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Prana A股财务分析助手" (gocybertrade/prana-stock-scoring-analysis) from ClawHub.
Skill page: https://clawhub.ai/gocybertrade/prana-stock-scoring-analysis
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install prana-stock-scoring-analysis

ClawHub CLI

Package manager switcher

npx clawhub@latest install prana-stock-scoring-analysis
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
medium confidence
Purpose & Capability
Name/description, required env var PRANA_SKILL_API_FLAG, and all declared network endpoints (https://claw-uat.ebonex.io) align: the skill simply proxies questions to a remote Prana/Claw agent and reads/writes a thread_id file for session continuity.
Instruction Scope
SKILL.md restricts activity to (a) calling the listed claw-uat.ebonex.io endpoints, (b) reading/writing workspace/<session_id>/prana-stock-scoring-analysis.txt for thread_id, and (c) requesting a PRANA_SKILL_API_FLAG token. It does not ask to read unrelated files or other env vars. Notes of concern: the skill requires the host/operator to create a global persistent env variable and to perform the workspace file read/write manually (the included client scripts do not write the thread file). The manual file operations are unusual and increase chance of operator error.
Install Mechanism
There is no install spec — included are only two simple client scripts (JS/Python). No downloads from third-party URLs or archives. This is a low-risk, instruction-only install footprint.
Credentials
Only one credential (PRANA_SKILL_API_FLAG) is required, which matches the declared auth header usage. However SKILL.md instructs persisting this token as a global OpenClaw env var, meaning the credential will be reusable by other skills or agent runs. If the token is shared or long-lived, that increases blast radius — prefer short-lived or per-session tokens where possible.
Persistence & Privilege
The skill does not request always:true and does not modify other skills. The main persistence request is instructing the operator to set a global env var and to persist a per-session thread_id file. Combined with normal autonomous invocation capability, a stored API key could be used by future autonomous runs — consider this when granting the env var.
Assessment
This skill is internally consistent with its description: it sends user questions to a remote Claw/Prana agent and keeps a per-session thread_id in workspace/<session_id>/prana-stock-scoring-analysis.txt. Before installing, verify you trust the endpoint https://claw-uat.ebonex.io (it appears to be a UAT/test host and no homepage/source is provided). Do not blindly persist secrets: if you must set PRANA_SKILL_API_FLAG, prefer an explicitly scoped, revocable token and confirm how to revoke it. Because the skill asks you to store the key as a global env variable, any other skill or autonomous agent run could reuse it — use an isolated environment or ephemeral token for testing. Also confirm your workspace path handling is safe (the scripts do not write the thread_id file for you) and ensure operators follow the SKILL.md instructions precisely to avoid leaking the token or mis-associating session files. If you need higher assurance, ask the publisher for production domain documentation, token semantics (shared vs per-user), and a homepage or source repository before deploying in production.
scripts/prana_skill_client.js:111
Environment variable access combined with network send.
Confirmed safe by external scanners
Static analysis detected API credential-access patterns, but both VirusTotal and OpenClaw confirmed this skill is safe. These patterns are common in legitimate API integration skills.

Like a lobster shell, security has layers — review code before you run it.

latestvk973gz3zx22s0tqq331mbp1g8n840xgs
129downloads
0stars
5versions
Updated 3w ago
v1.0.4
MIT-0

当客户提出需要对指定A股股票的股财务分析时,可以使用本技能。 例如:帮我分析茅台股票的盈利能力、帮我分析中国平安的营运能力

一、获取标识:从 Prana 平台获取 api_key(需先与用户郑重确认,且以用户触发运行为准配置环境变量)

  1. 检查当前 OpenClaw 环境中是否已配置环境变量 PRANA_SKILL_API_FLAG
  • 若环境变量已存在,直接进入运行流程;

  • 若任一环境变量不存在,执行第2步获取prana平台请求标识。

  1. 调用接口 GET /api/v2/api-keys,获取 api_key 的值(本标识仅用于 Prana 平台相关接口,不作用于任何其他数据)。

接口调用命令:

curl -sS "https://claw-uat.ebonex.io/api/v2/api-keys"

接口正常响应示例:

{
  "code": 200,
  "message": "success",
  "data": {
    "api_key": "af:XXXXX"
  }
}
  1. 将接口返回的 api_key 配置到 OpenClaw 环境变量:
  • 接口返回的api_key → 环境变量PRANA_SKILL_API_FLAG;

补充说明(OpenClaw环境变量配置规范):

需通过OpenClaw专属配置命令设置环境变量,确保配置全局生效:

openclaw config set env.PRANA_SKILL_API_FLAG "af_..."

二、运行流程:调用 Prana 平台接口获取数据

OpenClaw 执行者须知(必须遵守,否则视为未完整执行本技能)
本技能依赖「会话目录 + 文本文件」保存 thread_id客户端脚本不会自动创建或写入 workspace/<session_id>/。必须由执行本技能的 OpenClaw/宿主在跑脚本之前读文件、在拿到接口结果之后写文件。

  1. 检查当前 OpenClaw 环境中是否已配置环境变量 PRANA_SKILL_API_FLAG
  • 若未检测到环境变量,提示“环境变量不存在,检查环境变量是否生效”,并终止整个流程;

  • 若环境变量检测成功,进入第2步。

  1. 必须先解析会话 ID,再跑脚本
    • 必须从 OpenClaw 当前会话上下文取得 <session_id>(与 workspace/<session_id>/ 目录一一对应;若平台以其他字段命名,仍以实际会话唯一标识为准)。不得在未确定 <session_id> 的情况下假定路径。
    • 必须保证目录 workspace/<session_id>/ 存在:若不存在则先创建该目录,再执行后续读写。
    • 脚本参数 question:用户要执行的需求任务(例如:帮我分析茅台股票的盈利能力、帮我分析中国平安的营运能力)。不要包含账号密码、付款信息、个人隐私等敏感内容。
    • 脚本参数 thread_id必须先尝试读取 workspace/<session_id>/prana-stock-scoring-analysis.txt整文件内容(去首尾空白)作为 -t 的值;仅当文件不存在、为空或读取失败时,-t 才传空(或不传 -t)。
    • 根据本机已安装的客户端选择运行方式,默认优先使用 Node.js;若未安装 Node.js 或无法使用时再选用 Python 3
    • 执行脚本后必须立即执行步骤 3;仅执行脚本而不维护 prana-stock-scoring-analysis.txt 不算完成本技能。
  • Node.js 18+(默认)

    NODE_OPTIONS=--experimental-default-type=module node scripts/prana_skill_client.js -q "帮我分析茅台股票的盈利能力" [-t thread_id]
    
  • Python 3(备选) python3 scripts/prana_skill_client.py -q "帮我分析茅台股票的盈利能力" [-t thread_id]

  1. 必须写回 thread_id(与步骤 2 同一会话)
    • 使用与步骤 2 相同<session_id>
    • 从步骤 2 的脚本/接口响应中解析出 thread_id;若成功取得非空 thread_id必须将其仅写入(覆盖写入即可)workspace/<session_id>/prana-stock-scoring-analysis.txt,文件内容仅为该字符串,不要附加说明或 JSON。
    • 若本次响应中无有效 thread_id,则不要删除已有文件(保留上次值便于续聊),除非业务明确要求新开会话(此时应清空或删除该文件后再传空 -t)。

三、获取历史请求地址

用于获取可在浏览器中打开的 历史请求(技能获取记录) 页面链接。

  1. 检查当前 OpenClaw 环境中是否已配置环境变量 PRANA_SKILL_API_FLAG
  • 若未检测到环境变量,提示“环境变量不存在,检查环境变量是否生效”,并终止整个流程;

  • 若环境变量检测成功,进入第2步(调用接口)。

  1. 调用接口 GET /api/claw/skill-purchase-history-url
  • 成功时:从响应体 data.url 取出链接。不要把返回的完整链接写进日志以及存储;把完整链接直接发给用户即可。

接口调用命令(成功时返回执行结果 JSON;请求头中的x-api-key从环境变量PRANA_SKILL_API_FLAG获取):

curl -sS -H "x-api-key:af..." "https://claw-uat.ebonex.io/api/claw/skill-purchase-history-url"

接口正常响应示例:

{
  "code": 200,
  "message": "success",
  "data": {
    "url": "https://claw-uat.ebonex.io/skill-purchase-history-url?pay_token=xxxxxxx"
  }
}

四、安全与审计说明

  1. 本技能的远程执行步骤只调用 https://claw-uat.ebonex.io 的 claw agent 接口(例如 POST /api/claw/agent-run)。
  2. 为什么必须调用远程 https://claw-uat.ebonex.io?只有https://claw-uat.ebonex.io中才提供了完成本技能分析所需要的相关数据源和agent。
  3. 为什么必须写入“全局持久化环境变量”?持久化变量的目的之一是减少重复握手/调用次数,从而尽可能降低对平台每日资源消耗的影响(具体策略以平台实际计费/配额为准)。
  4. prana服务公网首页地址:https://claw-uat.ebonex.io/
  5. 关于 /api/v2/api-keys 的授权语义:不同环境/平台策略可能导致该接口在当前调用方式下无需额外认证。接口返回的令牌类型(共享令牌或用户专属令牌)及具体原因,请以平台文档/接口说明为准,以满足审计要求。

Comments

Loading comments...