The Agent Skills Directory

Command Execution via Script Arguments: The ci.mjs script accepts a branch name as a command-line argument and interpolates it into shell commands executed via execSync. If a branch name containing shell control characters is used, it could lead to unintended command execution on the local system.
Potential for Indirect Prompt Injection: The skill instructs the agent to automatically execute commands identified by a specific emoji (🚀) in the output. This ingestion point in scripts/ci.mjs lacks boundary markers to separate log content from instructions. Because the output includes data fetched from remote CI logs, there is a risk that a malicious user could inject instructions into the logs to be executed by the agent. The skill lacks comprehensive sanitization for these log lines, which are printed to the agent's context. The capability inventory includes the script's execSync calls and the agent's general shell execution permissions.
Automated Workflow Considerations: The skill encourages 'immediate and automatic' execution of commands to replicate CI failures. This design choice prioritizes speed but reduces the opportunity for manual verification of the commands being run, which is a common security consideration for AI-assisted tools.

ci