soia-dev-task-execute
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation includes an installation command using
npxto addsoia-team/soia-open-dev-coding-skills. This resource belongs to the author's namespace and is used for skill management within the platform. - [COMMAND_EXECUTION]: The workflow requires the agent to execute arbitrary shell commands for task verification, including unit tests, integration tests, and build checks. This is a functional requirement of the skill but involves executing code from the target workspace.
- [PROMPT_INJECTION]: The skill is subject to Indirect Prompt Injection (Category 8) risks. It ingests untrusted data from project workspaces, such as source code, configuration files, and documentation. Malicious instructions embedded in these files could attempt to influence the agent's behavior during the implementation or verification stages.
- Ingestion points: Reads files and environment details from the target workspace (SKILL.md).
- Boundary markers: Instructions specify defining a 'verification plan' and 'boundary' (Target vs Non-target) to mitigate scope creep.
- Capability inventory: Perform file writes, shell command execution (subprocess), and workspace reads (SKILL.md).
- Sanitization: Includes explicit instructions to avoid printing credentials or private information in completion receipts.
Audit Metadata