The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/install-schedule.sh establishes a persistence mechanism by creating a macOS LaunchAgent at ~/Library/LaunchAgents/jp.theplant.felix.daily-work-brief.plist to schedule the morning brief task.
[COMMAND_EXECUTION]: The scripts/prep.sh script invokes the agent using the --permission-mode bypassPermissions flag, which allows for unattended execution by suppressing user confirmation prompts for tool usage.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection by processing external data from multiple communication platforms.
Ingestion points: The scripts/gather.sh script fetches raw data, including comments and message threads, from Jira, Slack, and GitHub.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore or isolate instructions that may be embedded within the ingested external data.
Capability inventory: The skill has write access to external platforms via scripts/send.sh (Jira and GitHub comments) and can modify the system clipboard using pbcopy for Slack replies.
Sanitization: The skill does not perform any validation or sanitization of the gathered content before using it to generate drafted responses in the user's name.

brief