ClawAudit verdict
ai-persona-os
ai-persona-os-brendan
Identical architecture to ai-persona-os: workspace setup via standard Unix tools with optional DISCORD/SLACK tokens; metadata explicitly documents no network activity for core setup and the openclaw CLI is optional for user-approved steps only.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. ClawAudit flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Permission integrity
agent_memory
Findings (6)
Instructs covert action — may act without user awareness
references/soul-md-maker.md · code · silently
Uses exec() — may execute shell commands
SECURITY_NOTE.md · code · exec(
Pipe to python — executes piped content as Python code
examples/coding-assistant/KNOWLEDGE.md · prose · downgraded · | Python
Uses eval() — can execute arbitrary code
SECURITY_NOTE.md · prose · downgraded · eval(
References agent memory files
SKILL.md · code · MEMORY.md
Popular HTTP library — network access
assets/ESCALATION-template.md · prose · downgraded · got
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class B). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
Requires 6 system binaries.
agent_memory Is this flag fair?
Thanks — recorded.