
from agentsociety1,083
Create, validate, package, and upload datasets to the AgentSociety platform following the project's dataset schema and upload workflow.
This skill helps users build compliant dataset packages for AgentSociety by scaffolding directory structure, validating metadata, packing ZIP archives, and uploading/submitting datasets for review. It enforces id constraints and category choices and supports large uploads via multipart flows. Typical use: researchers or developers preparing experiment outputs, agent profiles, or survey data for sharing.
Use when you need to publish or share a dataset on AgentSociety: initializing a dataset repo, validating a dataset directory or ZIP, packing for distribution, or uploading and submitting for admin review. Avoid using this skill for browsing/downloading datasets.
Compatible with agents that can run Python tooling and CLI scripts (agents with Python/CLI support, e.g., Codex/CLI-capable agents).
A CLI tool for creating, validating, packaging, and uploading datasets to the AgentSociety platform. The script is pure stdlib Python with no external dependencies. It ran without import errors but exited with code 1 when invoked without a subcommand (expected behavior — shows help). The SKILL.md is well-structured with clear triggers, workflow diagrams, common mistakes table, and format specification. No security concerns found: no hardcoded secrets, no destructive commands, no shell injection risks, and credentials are stored with proper file permissions.
Clean, well-structured skill. Pure stdlib implementation — no pip dependencies needed. The script implements a complete dataset lifecycle: auth (Device Code Flow with token refresh), init, validate, pack, multipart upload, and submit. The SKILL.md follows good practices with specific trigger conditions, anti-triggers, and a workflow diagram. Narrow usefulness (AgentSociety platform only) but high quality within that scope.