ClawAudit verdict
dream-selfimproving
The skill performs nightly self-modification — auto-generating new skills from gap analysis and writing them to ~/SharedSkills/, editing cron jobs via 'openclaw cron edit --confirm', and pruning low-frequency skills — which is unusually broad self-modifying authority for an agent memory tool, even though the behavior is transparently documented.
⚠ 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 (5)
Pipe to python — executes piped content as Python code
references/diagnostic-checklist.md · code · | python3
Writes to SKILL.md — self-modifying skill
scripts/dream.py · prose · downgraded · SKILL.md").write
References agent memory files
SKILL.md · code · MEMORY.md
subprocess execution — runs system commands from Python
check_skill.py · prose · downgraded · subprocess.run(
Python os.environ.get — reads environment variable
bundle-search.py · prose · downgraded · os.environ.get(
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class D). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
Requires 1 system binary.
agent_memory Is this flag fair?
Thanks — recorded.