The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to clone an external repository from an untrusted source (https://github.com/Eng0AI/happiness-dashboard-template.git) and immediately proceed to install dependencies (pnpm install) and run project scripts (pnpm run sources). This sequence allows code hosted on the external repository to execute arbitrary commands with the agent's privileges.
[EXTERNAL_DOWNLOADS]: The skill downloads a complete project codebase from a GitHub account ('Eng0AI') that is not recognized as a trusted organization or well-known service provider.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to perform sensitive shell operations, including recursive directory deletion (rm -rf), moving files (mv), and initialization of new version control repositories (git init).
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection via its data processing workflow.
Ingestion points: The skill reads and processes data from CSV files located in the sources/happiness_score/ directory.
Boundary markers: There are no specified delimiters or instructions for the agent to ignore potentially malicious content embedded within these data files.
Capability inventory: The skill environment has the capability to execute shell commands through pnpm and deploy content to external providers like Vercel and Netlify.
Sanitization: No sanitization or validation logic is defined to check the contents of the ingested CSV data before it is processed by the Evidence framework or used in visualizations.

happiness-dashboard