{"slug":"strategy/business-strategy/31-governance","title":"Governance & Approval","section":"strategy","sectionLabel":"Strategy","status":"unclassified","type":"document","path":"strategy/business-strategy/31-governance.mdx","bytes":2599,"description":null,"updated":null,"source":"integratewise/Integratewise-docs-wiki","content":"\nGovernance is the hard boundary between proposal and execution. IntegrateWise does not allow consequential actions to bypass governance.\n\nFor the canonical governance model, see `docs/spec/INTEGRATEWISE_CANONICAL_SPEC_v1.0.md`, `docs/architecture/UNIFIED_GOVERNANCE_SPEC.md`, and `docs/platform-specs/09-governance-engine.md`.\n\n## Two gates\n\nGovernance uses two phases:\n\n### Pre-Proposal Gate\n\nEvaluates:\n\n- authority\n- scope\n- policy\n\nPossible results:\n\n- ALLOW_PROCEED\n- REQUIRE_PRE_APPROVAL\n- DENY\n\nA 0.99 confidence proposal with missing authority is still DENY.\n\n### Post-Proposal Gate\n\nEvaluates:\n\n- risk\n- evidence\n- confidence\n- approval requirement\n\nPossible results:\n\n- AWAITING_HUMAN_APPROVAL\n- HUMAN_APPROVED\n- DENY\n\nConfidence never grants authority. High confidence does not remove approval requirements for consequential actions automatically.\n\n## Approval Center\n\nThe Approval Center is the user-facing governance surface. Every proposed consequential action must flow through it.\n\nApproval records include:\n\n- proposed action\n- business purpose\n- initiating actor\n- affected contexts and systems\n- exact fields or operations\n- evidence\n- execution-plan version\n- risk classification\n- approval policy\n- expected outcome\n- rollback or recovery availability\n\nAvailable decisions:\n\n- Approve\n- Reject\n- Request Changes\n- Edit and Approve when policy permits it\n- Delegate Approval\n- Cancel\n\n## ExecutionPlan contract\n\nThe user never approves an AI message. The user approves an ExecutionPlan.\n\nExecutionPlan contains:\n\n- plan identity and tenant scope\n- proposal lineage\n- immutable plan version\n- summary and impact summary\n- step dependencies and compensation steps\n- affected entities and systems\n- authority and evidence references\n- execution mode and sync mode\n- rollback strategy\n\nAfter approval, the Workbench shows execution progress, provider responses, partial completion states, failure and retry states, sync results, and final outcome.\n\n## Governing rules\n\n- Approval Center is the only surface that can mint an approval token.\n- No handoff package can be generated without approval lineage.\n- Twin does not auto-execute.\n- Governance rules are compiled configurations. They cannot be modified by model outputs or agent workflows.\n\n## Audit expectations\n\nEvery consequential action must log:\n\n- actor\n- authority scope\n- policy validation result\n- affected context\n- exact provider operations\n- execution-plan version\n- approval record\n- provider outcome\n- sync result\n- tamper-proof audit event\n\nThis is not optional in production integrations.\n"}