memory-management

Pass

Audited by Gen Agent Trust Hub on Apr 28, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through its data ingestion pipeline.
  • Ingestion points: The agent is instructed to read context from files within the memory/ directory, including memory/hot-cache.md, memory/glossary.md, and archived data in memory/archive/. If these files are populated with untrusted content (e.g., from a compromised SEO data source), they could contain hidden instructions.
  • Boundary markers: There are no explicit instructions to treat content from these files as untrusted or to wrap them in boundary markers (like XML tags) to separate data from system instructions.
  • Capability inventory: The skill possesses significant capabilities, including writing to the file system, managing project structure, and coordinating with other skills that may perform network operations (like rank-tracker or seo-content-writer).
  • Sanitization: The instructions lack requirements for sanitizing or validating the content retrieved from the memory files before using it to determine the agent's next actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 28, 2026, 04:12 PM