plugin-forge
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides utility scripts (
create_plugin.py,bump_version.py) and documentation for managing local plugin projects. These tools facilitate directory creation, template generation, and JSON manifest updates, all of which are standard development tasks without any unauthorized access or network communication. - [INDIRECT_PROMPT_INJECTION]: The skill features scripts that ingest metadata (such as author names, descriptions, and keywords) via command-line arguments to generate and update plugin manifest files (
plugin.json,marketplace.json). This creates a surface where untrusted data could be injected into project files; however, this is the primary intended function of the tool and is handled using standard JSON serialization which prevents basic structural breakage. - Ingestion points: CLI arguments in
scripts/create_plugin.py(description, author, keywords) andscripts/bump_version.py(plugin names). - Boundary markers: Absent.
- Capability inventory: File system write operations within the designated marketplace root directory.
- Sanitization: Absent (the scripts perform direct JSON dumping of the provided input).
Audit Metadata