tasks
Pass
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
Bashtool to executealmaCLI commands for creating, listing, and updating tasks stored in the local filesystem. - [PROMPT_INJECTION]: The skill architecture allows for indirect prompt injection because task metadata is shared globally and automatically injected into the agent's context in every conversation thread.
- Ingestion points: Task data is retrieved from
~/.config/alma/tasks.jsonand injected into the agent context. - Boundary markers: There are no documented delimiters to segregate injected task content from system instructions.
- Capability inventory: The skill is configured with access to the
Bashtool, which provides a significant capability for system interaction. - Sanitization: No sanitization or validation of user-provided task strings is described before they are processed by the agent.
Audit Metadata