禅道自动报告

禅道自动记工时技能。用户描述今天做了什么事，自动搜索匹配任务，确认后调用禅道API记录工时。支持智能匹配、自动重新登录、正确计算剩余工时。使用 recordworkhour 接口，适配禅道 21.x 开源版。

This skill appears to implement what it claims, but review a few risks before installing: - Credentials handling: You must provide account/password (or token) and the skill expects them in plaintext at $HOME/.config/zentao/.env. If you proceed, restrict that file's permissions (chmod 600) and consider using an API token instead of a password where possible. - Shell-script env loading: The bundled .sh scripts use 'export $(cat <file> | xargs)' to load the .env. That approach can evaluate command substitutions if the file contains crafted values; ensure you create the .env yourself (do not paste untrusted content) and inspect it before use. - Command/argument interpolation: The scripts embed user-provided descriptions into shell/python invocations. If you plan to run these scripts on a shared or untrusted machine, validate inputs to prevent accidental injection. Prefer the Python scripts over the shell ones when possible (they are easier to audit and avoid shell word-splitting quirks). - Endpoint trust: The code will POST workhours to whatever ZENTAO_URL you supply. Double-check that URL (use HTTPS) to avoid sending credentials/work entries to a wrong or malicious server. - Data exposure: The matching step prints full task lists and JSON; this will surface internal task metadata into the agent/UI. If that is sensitive, consider restricting output or running locally. If you are comfortable with these caveats (and/or can run the scripts in a local, trusted environment and set secure file permissions), the skill is functionally coherent. If you have limited ability to enforce file permissions or are concerned about credential storage, do not install or require using an API token and manually inspect and run the Python scripts instead of the shell variants.

Type: OpenClaw Skill Name: zentao-autoreport Version: 1.0.2 The skill automates Zentao work reporting but contains several security vulnerabilities. Specifically, scripts/match-tasks.sh and scripts/report.sh are vulnerable to code injection because they embed unsanitized shell variables (like $USER_DESC) directly into Python one-liners executed via 'python3 -c'. Additionally, the skill stores Zentao credentials in plaintext in $HOME/.config/zentao/.env and session cookies in /tmp/cookies.txt, which are insecure practices. While these appear to be unintentional flaws rather than intentional malware, they represent significant security risks.