rlm-distill

Pass

Audited by Gen Agent Trust Hub on Mar 19, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill uses role-play and override instructions to influence the agent's behavior: 'You ARE the distillation engine. You replace the local Ollama distiller.py script' and 'Never run distiller.py'.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface:
  • Ingestion points: The agent reads the 'entire file' using the view_file tool for distillation.
  • Boundary markers: Absent; the agent is not provided with delimiters or instructions to ignore embedded commands within the files it reads.
  • Capability inventory: The agent has access to the Bash tool (command execution) and Write tool, along with the ability to execute Python scripts.
  • Sanitization: Absent; no input validation or filtering is specified before the agent processes file content.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute multiple local Python scripts for inventory, data injection, and batch processing: 'inventory.py', 'inject_summary.py', and 'swarm_run.py'.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 19, 2026, 11:28 PM