compound

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill extracts information from the active conversation to create persistent memory files, creating a surface for indirect prompt injection.
  • Ingestion points: Conversations and session history (SKILL.md).
  • Boundary markers: Absent. The skill does not use specific delimiters or instructions to ignore embedded commands within the extracted learnings.
  • Capability inventory: The skill has file-writing and file-editing capabilities, including the ability to delete or replace text in memory files using shell commands like grep and bash.
  • Sanitization: Absent. The skill does not explicitly sanitize or validate the content extracted from the session before writing it to persistent storage.
  • [COMMAND_EXECUTION]: The skill uses shell commands (bash, grep) to manage memory files, check for contradictory entries, and verify that global bridges are correctly established. These operations are limited to the user's home and project directories.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 11:19 PM
Security Audit — agent-trust-hub — compound