groove-utilities-memory-promises
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes various subcommands using the
beansCLI tool (e.g.,beans list,beans create,beans update) through a scoped bash environment (Bash(beans:*)). These commands are used to manage milestones, epics, and tasks within the user's workspace. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from
.groove/index.mdand command outputs from thebeanstool. - Ingestion points: Configuration is read from
.groove/index.md, and task lists are retrieved from thebeansCLI output. - Boundary markers: The instructions do not define clear delimiters or "ignore instructions" headers for the data processed from files or tool outputs.
- Capability inventory: The skill has the capability to write/edit files and execute bash commands via the
beanstool. - Sanitization: There is no evidence of explicit sanitization or validation of the text retrieved from the configuration file or the task backend before it is used in subsequent operations.
Audit Metadata