All autonomous operations in strict mode must execute through this wrapper:
python3 /home/openclaw/.openclaw/workspace/governance/governance_wrapper.py
Required parameters:
--requested-skill--system-prompt--input-context
Strict-mode controls enforced by the wrapper:
- Model lock (
opencode/big-pickle, no fallback, temperature 0.0) - Skill allowlist enforcement from tool surface manifest
- Network allowlist enforcement for outbound HTTP
- Subagent semaphore (
maxConcurrentSubagents) - Mandatory
execution-evidence.v1emission - Hash-chained append-only evidence logging
Any policy violation results in a blocked execution with recorded evidence.