伐谋 - 数据分析工具
v1.0.0数据分析技能，用于理解数据、分析数据、制作数据处理流程、汇总数据分析结果。当用户提到"分析数据"、"数据处理"、"数据探索"、"统计分析"、"数据清洗"、"数据汇总"、"制作数据报告"、"理解这份数据"、"看一下这个CSV/Excel/数据集"时，必须使用此技能。即使用户只说"帮我看看这个数据"、"分析一下"，只...
⭐ 0· 574·1 current·1 all-time
byFaMou@zhaom0
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (数据分析) matches the instructions: guidance on understanding, cleaning, transforming, and reporting on datasets. No unrelated credentials, binaries, or installs are requested.
Instruction Scope
Instructions stay within data-analysis tasks (quality checks, cleaning rules, reporting format). It explicitly references local upload/output paths (/mnt/user-data/uploads/ and /mnt/user-data/outputs/) and runtime settings (encoding, matplotlib fonts), which is reasonable for a data skill but worth noting because the skill assumes access to those filesystem locations and will be triggered whenever the user mentions data analysis.
Install Mechanism
No install spec and no code files — instruction-only skill. This minimizes on-disk footprint and external downloads.
Credentials
Requires no environment variables, credentials, or config paths. The only environment assumptions are file locations and common encoding/font settings needed for data processing and visualization.
Persistence & Privilege
always is false and the skill does not request persistent system modifications or elevated privileges. Autonomous invocation is allowed by platform default but not combined with other high-risk factors.
Assessment
This skill is coherent for dataset analysis: it doesn't ask for credentials or install code. Before enabling, confirm that your agent environment uses the referenced paths (/mnt/user-data/uploads/ and /mnt/user-data/outputs/) and that you are comfortable the agent can read uploaded files placed there. Also note the skill declares it should be invoked whenever a user mentions data-analysis terms — consider whether you want it auto-triggered in all those cases. Finally, as with any data-processing tool, avoid uploading sensitive data unless you trust the agent/environment's storage and retention policies.Like a lobster shell, security has layers — review code before you run it.
latestvk978hz9v6549s3shc3cpn2f9p982vg8e
License
MIT-0
Free to use, modify, and redistribute. No attribution required.