The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard shell commands like grep and git log to search local files and history. These operations are scoped to the project's own metadata and do not pose a security risk in this context.- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local .trail/ files. While these files are technically external input to the agent's context, the skill is limited to reading and summarizing this information for the user, with no dangerous downstream capabilities like file writing or network access.- [SAFE]: No suspicious patterns, obfuscation, or data exfiltration attempts were detected.

trail-why