The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/render_wechat_2d.sh clones an external repository (https://github.com/sxhzju/wechat-2d.git) and proceeds to execute its contents. Specifically, it runs pnpm install, which can trigger post-install scripts, and pnpm run remotion:render, which executes project-specific build and rendering logic.
[EXTERNAL_DOWNLOADS]: The skill relies on git clone to fetch the primary application logic from a third-party GitHub repository. It also uses pnpm to download and install numerous Node.js dependencies from public registries during the setup process.
[COMMAND_EXECUTION]: The workflow involves multiple shell command executions, including corepack enable pnpm, git operations, and pnpm script runners. These commands are used to manage the environment and drive the rendering engine.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill ingests a user-provided props_file (JSON) and content from the external wechat-2d repository during the render process.
Boundary markers: No explicit delimiters or instructions are used to prevent the agent from interpreting data inside the props_file or the external repository as instructions.
Capability inventory: The skill has the capability to write to the file system (mkdir, git clone), execute shell commands, and perform network operations via git and pnpm.
Sanitization: The script uses standard shell quoting for variables, but does not perform deep validation or sanitization of the JSON properties passed to the Remotion engine.

wechat-2d-render