Code Refactor for Reproducibility
v1.0.0Use when refactoring research code for publication, adding documentation to existing analysis scripts, creating reproducible computational workflows, or prep...
⭐ 0· 33·0 current·0 all-time
byAIpoch@aipoch-ai
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description describe refactoring research code. The package contains an analysis/refactor entrypoint (scripts/main.py) and README/template generation consistent with that purpose. No unrelated binaries, services, or credentials are requested.
Instruction Scope
SKILL.md instructs the agent to analyze source files, confirm inputs/outputs, run the packaged Python script, and produce reproducible outputs. Those instructions remain within the stated scope and explicitly recommend backups and scope confirmation before running.
Install Mechanism
There is no install specification — instruction-only skill with an included Python script. No external downloads, package installs, or extraction from arbitrary URLs are present in the metadata.
Credentials
The skill requests no environment variables or credentials. The included requirements.txt lists common scientific packages; one entry 'src' is unusual (may reference a local package) but not a credential or unrelated secret.
Persistence & Privilege
always is false and the skill does not request persistent platform privileges. The script will create files/directories in the chosen output path (expected for this purpose) but does not modify other skills or system-wide agent settings.
Assessment
This skill appears coherent and self-contained, but it executes a local Python script that will read source files and write a project structure. Before running: (1) inspect scripts/main.py locally (search for network calls or subprocess.exec if you want extra assurance), (2) run python -m py_compile scripts/main.py and python scripts/main.py --help as suggested, (3) run it in an isolated directory or sandbox and back up originals to avoid accidental overwrites, and (4) note requirements.txt includes a 'src' entry which may reference a local package—verify your environment before pip/conda installs. If you will allow autonomous agent invocation, remember the agent could run the script; only grant that if you trust automated runs.Like a lobster shell, security has layers — review code before you run it.
latestvk97232ayjcks614dw0z0vzyy9d843chb
License
MIT-0
Free to use, modify, and redistribute. No attribution required.