The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructs the agent to modify its own source code ("Self-Evolving Skill", "fix this file immediately"). This allows the agent to permanently alter its own instructions, creating a risk of persistent behavioral change.
[COMMAND_EXECUTION]: The skill uses bash to perform system service management tasks, including launchctl bootout and file deletion of .plist configuration files. It also dynamically sources script files from paths determined by environment variables.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface.
Ingestion points: Reads sensitive user transcripts from ~/.claude/projects (SKILL.md).
Boundary markers: None present; the skill does not use delimiters or instructions to ignore embedded commands in the transcripts.
Capability inventory: The skill can execute arbitrary shell commands via bash (SKILL.md) and modify its own source code.
Sanitization: No sanitization or validation of the transcript content is performed.
Risk: Malicious instructions embedded in user transcripts could trigger the agent to modify its own source code with harmful logic.

doctor