slash-command-creator
Warn
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION] (MEDIUM): The skill facilitates the creation of files that enable bash command execution via the
!prefix. Although it documents a security feature (allowed-tools) to restrict patterns, the capability itself presents a risk if misconfigured or if a malicious command is generated by the agent under duress. - [PERSISTENCE] (MEDIUM): The
scripts/init_command.pyscript is designed to write executable-like configuration files to~/.claude/commands/. This allows instructions and tool-use permissions to persist across sessions for the host application. - [INDIRECT_PROMPT_INJECTION] (LOW): The slash command architecture relies on interpolating untrusted data ($ARGUMENTS and @file references) directly into prompts.
- Ingestion points:
$ARGUMENTSand@filemarkers inSKILL.mdandreferences/examples.md. - Boundary markers: Absent; examples show direct interpolation without delimiters or escaping instructions.
- Capability inventory:
scripts/init_command.py(file-write), and documentedBashtool usage. - Sanitization: None provided in the skill scripts; the skill assumes the host environment handles sanitization.
Audit Metadata