wasp-plugin-init
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is designed to initialize a Wasp development environment for AI agents. It does not contain any malicious patterns such as credential exfiltration, persistence backdoors, or privilege escalation.
- [COMMAND_EXECUTION]: The skill uses standard shell commands (
rm,mkdir,cp,touch) to manage versioning and local state in the.claude/wasp/directory. These operations are transparent, scoped to the project, and required for the skill's stated purpose of maintaining plugin version state. - [EXTERNAL_DOWNLOADS]: The instructions direct the agent to fetch documentation from
wasp.sh(the official vendor domain) andraw.githubusercontent.com. These are trusted sources for project documentation and do not pose a security risk. - [INDIRECT_PROMPT_INJECTION]: The skill updates the agent's persistent memory (
CLAUDE.mdorAGENTS.md) with instructions and best practices. While this modifies future agent behavior, the process is initiated only after explicit user confirmation (AskUserQuestion) and uses local content provided within the skill package (general-wasp-knowledge.md).
Audit Metadata