skills/antjanus/skillbox/remember/Gen Agent Trust Hub

remember

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Executes local git commands including git log, git status, and git stash list to reconstruct recent work history and current repository state.
  • [PROMPT_INJECTION]: The skill processes untrusted data from previous conversation logs and project files, creating an indirect prompt injection surface.
  • Ingestion points: Reads session history JSONL files from ~/.claude/projects/, and project-level files such as SESSION_PROGRESS.md and ROADMAP.md.
  • Boundary markers: No specific delimiters or boundary markers are defined to isolate untrusted session history content from the agent's internal instructions.
  • Capability inventory: Executes shell-based git commands and has file system read access.
  • Sanitization: The skill contains a mandatory rule to sanitize sensitive data (API keys, tokens, credentials) before producing the summary, but does not explicitly handle potential instruction injection within those logs.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 11:52 PM