superhuman
Pass
Audited by Gen Agent Trust Hub on Mar 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill includes a 'Companion check' feature that identifies missing recommended tools and offers to install them from the author's official repository via the
npx skills addcommand. - [COMMAND_EXECUTION]: To support its development workflow, the skill executes standard project scripts and development utilities including
git,npm,yarn,pytest,make, andtscfor build and test verification. - [DATA_EXFILTRATION]: The skill reads local project configuration files (e.g.,
package.json,pyproject.toml), environment metadata, and documentation to establish context for development tasks. It maintains persistent state in a.superhuman/board.mdfile within the project directory. - [PROMPT_INJECTION]: To mitigate risks when delegating tasks to parallel sub-agents, the skill utilizes a 'Agent Context Handoff Format' with structured Markdown delimiters to isolate user-provided context from executable instructions.
Audit Metadata