Conscience
Conscience is the policy engine and governance layer. It defines what your agents are allowed to do, records what they did, and emits tamper-evident receipts every paid op can be audited against.
Note
Included in: Starter, Pro, Team, Enterprise. Renders under
/conscience in Midcore Shell and the web dashboard.What Conscience does
- Declarative policy packs — tenant-scoped rule sets that gate egress, execution, identity, and data movement.
SYSTEM_RULESalways load first; tenant policies layer on top. - Fail-closed gates — if no policy explicitly allows an action, it’s denied. Every denial emits an audit row with the rule id.
- Hash-chained receipts — every cloud op produces a receipt linked to the prior one. Tamper detection is offline-verifiable.
- Jurisdictional templates — pre-built packs for GDPR, HIPAA, PIPEDA, SOC 2, ISO 27001, and Canadian + US tax-handling.
When to reach for Conscience
- You need an auditable record of who-did-what-when on AI actions.
- You operate in a regulated industry and need pre-built policy templates.
- You want to constrain agent behaviour (no egress to
X, no shell exec, etc.) declaratively.