build-task
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external, potentially attacker-controlled sources (such as issue descriptions or documentation) and interpolates it into prompts for other sub-agents. \n
- Ingestion points: Data is read from
docs/plans/*.mdfiles and external task references via theTask(majestic-engineer:workflow:task-fetcher)component. \n - Boundary markers: The workflow prompts do not employ specific delimiters (e.g., XML tags or unique separators) to encapsulate untrusted data, nor do they instruct sub-agents to ignore instructions embedded in the task content. \n
- Capability inventory: The skill has significant capabilities including shell command execution (
Bash), task creation and updates (TaskCreate,TaskUpdate), and the ability to trigger code generation and PR creation via thebuild-task-workflow-manager. \n - Sanitization: There is no explicit sanitization, filtering, or escaping of the external task content before it is passed to the planning and execution agents.
Audit Metadata