project-planning
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for ingesting project context from an external memory source, creating a surface for indirect prompt injection.
- Ingestion points: Data enters via memory tools mcp__memory__search_nodes and mcp__memory__open_nodes in SKILL.md.
- Boundary markers: None present; the instructions do not specify delimiters for fetched data.
- Capability inventory: The skill uses memory write tools (mcp__memory__create_entities, mcp__memory__add_observations, mcp__memory__create_relations) in SKILL.md to modify the knowledge base.
- Sanitization: Absent; no filtering of memory content is instructed.
Audit Metadata