softaworks-agent-toolkit
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download and install several specialized diagramming and architecture tools from the
softaworks/agent-toolkitrepository hosted on GitHub. GitHub is a well-known service, and the installation process is transparently handled through thenpx skillsutility. - [COMMAND_EXECUTION]: The skill provides the agent with shell commands to detect existing installations and execute the installation process via
npx. These commands are restricted to checking file presence ([ -f "$f" ]) and performing tool additions, with no attempts at privilege escalation or persistent access modification. - [INDIRECT_PROMPT_INJECTION]: By providing capabilities to read and process diagram data (such as Mermaid definitions, Excalidraw JSON, or Marp markdown), the skill introduces an indirect injection surface. Maliciously crafted diagram files could theoretically contain instructions intended to influence the agent's behavior, although this is a standard risk for tools processing external data formats.
Audit Metadata