{"id":"pattern.boundaries-increase-agency","slug":"useful-boundaries-increase-agency","type":"pattern","version":"0.1.0","maturity":"hypothesis","status":"published","title":"Useful boundaries increase agency","summary":"Define clear, enforceable limits and safe operating space so actors can move faster without repeatedly seeking permission.","audiences":["leader","security","agent-builder"],"relationships":{"supports":["pattern.accountability-authority","pattern.evidenced-trust"],"contradicts":[],"depends_on":["pattern.versioned-decisions"]},"disclosure":"green","approved_by":["chrisleeddotcom"],"license":"CC-BY-4.0","created":"2026-08-21","updated":"2026-08-21","canonical_url":"https://intentarium.com/patterns/useful-boundaries-increase-agency/","source":{"repository":"https://github.com/Intentarium/canon","commit":"a8b303818e72c20505bd9ef56f5fb2d0448aa6f3","path":"patterns/useful-boundaries-increase-agency.md"},"body_markdown":"## Context\n\nPeople or agents need autonomy, while the system must constrain material risk, cost, disclosure, or irreversible effects.\n\n## Problem\n\nVague control produces two bad outcomes: cautious actors stop for permission, while aggressive actors discover limits only after crossing them. More restriction then reduces legitimate agency without improving clarity.\n\n## Forces\n\n- Flexibility helps adapt to context.\n- Predictability enables independent action.\n- Controls can become obsolete.\n- Enforcement without explanation damages trust.\n\n## Pattern\n\nExpress boundaries as observable rules around a deliberately useful operating space. State what is allowed, prohibited, conditional, and escalated. Enforce close to execution, explain decisions, and provide a governed path for changing the boundary.\n\n## Apply it\n\nBegin with the outcome and material harms. Grant the widest reversible scope consistent with those harms. Provide safe defaults, budgets, rate limits, scoped tools, and explicit stop conditions. Review repeated escalation as a possible boundary-design defect.\n\n## Failure modes\n\n- The boundary names prohibited outcomes but has no enforcement.\n- Controls are so broad that routine work always escalates.\n- Exceptions are granted privately and cannot be learned from.\n- Actors optimize to the written rule while violating its intent.\n\n## Evidence needed\n\nOutcomes showing increased safe throughput and reduced ad hoc permission requests without higher loss, disclosure, cost, or rollback rates."}