crossmem-loader

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest external data which creates a surface for indirect prompt injection.
  • Ingestion points: The crossmem load --session <ref> --full command (SKILL.md) reads raw conversation transcripts from various agent tools into the current session.
  • Boundary markers: The skill mentions markers like [... middle of the session elided ...] and instructions to ignore harness boilerplate, but lacks robust machine-enforced delimiters between historical content and the current instruction stream.
  • Capability inventory: The skill provides access to several CLI operations including crossmem list, load, update, export, import, and config (SKILL.md, references/troubleshooting.md).
  • Sanitization: The instructions tell the agent to manually "filter noise" and synthesize a brief, which relies on the LLM's adherence to instructions rather than technical sanitization of the input.
  • [COMMAND_EXECUTION]: The skill requires the agent to execute a variety of subcommands through the crossmem CLI tool to list sessions, load context, and manage configurations.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 09:42 PM
Security Audit — agent-trust-hub — crossmem-loader