The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform destructive operations, including the removal of files via rm -f and the termination of processes using pkill -f. It also invokes a local shell script located at ~/eon/cc-skills/plugins/tts-tg-sync/scripts/kokoro-install.sh to perform uninstallation tasks.
[CREDENTIALS_UNSAFE]: The instructions explicitly target a secrets file at ~/.claude/.secrets/ccterrybot-telegram for deletion. Accessing or managing files within credential directories is a sensitive operation that carries risk if misused.
[PROMPT_INJECTION]: The skill includes a 'Post-Execution Reflection' mechanism that commands the agent to autonomously edit the SKILL.md file. This self-modification capability can be used to alter agent behavior or bypass security logic across subsequent sessions.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface through the AskUserQuestion tool, where untrusted user input is collected to influence component removal logic.
Ingestion points: User selection input via AskUserQuestion in Phase 1 of SKILL.md.
Boundary markers: None identified; the instructions do not specify delimiters for user-provided data.
Capability inventory: Subprocess execution via Bash (including rm, pkill, and local script execution).
Sanitization: None identified; the workflow lacks explicit validation or escaping of the user's choice before execution.

clean-component-removal