secondbrain-task
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input (task titles and descriptions) and writes this content into local files (
.claude/data/tasks/YYYY-MM.yamlanddocs/tasks/*.md). This creates an attack surface where a malicious user could embed instructions in a task description meant to influence the agent's behavior when it later reads or processes those task documents. - Ingestion points: User-provided strings for Title, Description, and Tags (SKILL.md).
- Boundary markers: The skill uses YAML structure for data storage, but does not specify delimiters or sanitization for the free-text description field.
- Capability inventory: File read/write access to project data directories.
- Sanitization: No explicit sanitization or instruction-filtering is mentioned for user-supplied task content.
Audit Metadata