Qa Scenario Synthesizer
v1.0.0把需求拆成功能、异常、跨端、脏输入和恢复场景测试集。；use for qa, test-scenarios, edge-cases workflows；do not use for 宣称已经完成测试, 忽略关键用户路径.
⭐ 0· 104·0 current·0 all-time
byvx：17605205782@52yuanchangxing
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description, SKILL.md, resources, and scripts all align: a local, read-only-focused QA/test-scenario generator and local audit helper. Declared requirement (python3) matches the bundled Python script; no unrelated cloud credentials, binaries, or config paths are requested.
Instruction Scope
Runtime instructions limit behavior to producing structured text and optionally running scripts/run.py. The script reads provided files/directories (including .py/.sh/.json/.md) to generate reports and pattern scans. This is expected for an auditing/generation tool, but means you should not point the skill at directories containing secrets or other sensitive files because it will read and may include snippets in outputs.
Install Mechanism
Instruction-only install (no remote download). The bundle includes a local Python script and resource files; there is no installer that fetches code from external URLs or executes remote archives.
Credentials
The skill requests no environment variables or credentials. Its functionality is consistent with this: it operates on local input files and templates only.
Persistence & Privilege
always is false, the skill does not request permanent/force-installed presence, and it does not modify other skills or global agent configuration. It may write requested output files (as expected) but otherwise has no elevated privileges.
Assessment
This skill appears to be what it says: a local QA/test-scenario synthesizer that uses a bundled Python script. Before running: (1) review scripts/run.py if you have concerns; (2) avoid pointing --input at directories containing secrets, private keys, or unrelated configs because the tool will read files and may include snippets in output; (3) run with --dry-run or point it at a copy of your data first; (4) if you plan to let an agent invoke this autonomously, be aware it may read files you supply — restrict inputs and run in an isolated workspace.Like a lobster shell, security has layers — review code before you run it.
latestvk973sw5gtdyykn5ahn4vjqj36s836s2y
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🧷 Clawdis
OSmacOS · Linux · Windows
Binspython3