session-logs

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes historical conversation data that may contain untrusted instructions from past sessions.
  • Ingestion points: The skill reads session logs from ~/.openclaw/agents/<agentId>/sessions/*.jsonl, which include past user messages, assistant responses, and tool results.
  • Capability inventory: Employs bash, jq, and rg to extract and display contents from these files into the agent's current context.
  • Boundary markers: The provided command templates (e.g., searching for keywords or listing sessions) do not implement delimiters or safety instructions to prevent the agent from being influenced by the retrieved text.
  • Sanitization: There is no evidence of content filtering or sanitization to identify or neutralize potential instructions embedded within the historical logs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 07:53 PM
Security Audit — agent-trust-hub — session-logs