relatr-elo-plugin-author

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONNO_CODE
Full Analysis
  • [NO_CODE]: The skill consists entirely of markdown documentation, templates, and reference materials. It does not include any executable scripts or binary files.
  • [COMMAND_EXECUTION]: The skill provides instructions for the user to execute CLI commands using the @contextvm/relo package (e.g., relo check, relo build, relo publish). These commands include placeholders for user-supplied secrets like the --sec flag for private keys.
  • [PROMPT_INJECTION]: The skill processes user-supplied Elo source code as part of its 'Review' and 'Debug' workflows, creating a surface for indirect prompt injection.
  • Ingestion points: User-provided Elo source code in 'Review' and 'Debug' tasks.
  • Boundary markers: Not present.
  • Capability inventory: The agent generates Elo code, manifest metadata (Nostr kind 765), and shell commands (relo).
  • Sanitization: No sanitization of the input Elo source code is performed before analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 09:25 PM