beans
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides setup instructions to download and install the Beans CLI tool from its official repository on GitHub (
github.com/hmans/beans) using common package managers like Homebrew and Go. It also references documentation from thecontext7.comdomain. - [COMMAND_EXECUTION]: The skill relies on executing the
beansCLI for core functionality, including initializing projects, listing task states, and performing targeted updates to task metadata and content via shell commands. - [INDIRECT_PROMPT_INJECTION]: The skill workflow involves the agent reading and acting upon data stored in task files within the
.beans/directory. These files are processed by the agent and could potentially contain untrusted instructions. - Ingestion points: Task content is ingested through CLI outputs from
beans showandbeans list --full, as well as direct reading of the.beans/*.mdfiles. - Boundary markers: No explicit delimiters or instructions to ignore nested prompts are specified for the bean file bodies.
- Capability inventory: The skill allows the agent to execute the
beansCLI, write files to the.beans/directory, and perform external web research for triage purposes. - Sanitization: The instructions do not specify sanitization or validation logic for the content processed from external bean files.
Audit Metadata