identity-guard

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local shell scripts (scripts/guard.sh, scripts/init.sh, scripts/add-user.sh, scripts/whoami.sh) to perform identity verification and configuration management. These scripts are part of the skill's distributed files.
  • [DATA_EXFILTRATION]: The scripts/whoami.sh utility reads session log files from the agent's directory (~/.openclaw/agents/main/sessions/) to help users identify their system ID. This involves accessing historical interaction data stored on the filesystem.
  • [COMMAND_EXECUTION]: The scripts/add-user.sh and tests/run.sh files execute Python code dynamically via shell heredocs (python3
  • <<'PY'). This code is used for structured data (JSON) manipulation and automated testing.
  • [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes sender_id and channel metadata from user messages.
  • Ingestion points: Message metadata fields described in SKILL.md.
  • Boundary markers: SKILL.md contains explicit instructions to ignore self-claimed identities and social engineering attempts, relying only on the verification script.
  • Capability inventory: Filesystem access to identities.json and session logs; execution of local scripts for authorization checks.
  • Sanitization: Scripts use Python's json module for safe data parsing and standard shell quoting for parameter handling.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 11:01 AM