Architecture & Security Guardrails¶
Enterprise plan feature
Organization guardrails are available on the Enterprise plan.
Guardrails are the architecture and security standards N45.AI respects across every discovery, spec, quick-feat and hot-fix in an organization. It's how the Owner ensures everything N45 plans and builds — in any project or workspace in the org — is born within the house rules.
Who sets them, and where¶
The organization Owner defines the guardrails in the N45 account (web), under Guardrails. Two lists, one item per line:
- Architecture standards — e.g. use Postgres (not Mongo); hexagonal architecture; always wrap errors; tests mandatory in the domain.
- Security guardrails — policies N45 must respect, including in quick-feat and hot-fix.
Guardrails are versioned: each change becomes a new version, with a history of who changed what and when.
Few rules, well chosen
Guardrails enter every execution. Too many rules increase token cost and dilute focus — prefer to consolidate into a lean set of standards that truly matter.
How N45 applies them¶
Guardrails are an always-on floor: they apply to every member of the org, with no one having to remember to turn them on. During planning (discovery and spec), N45:
- Adapts the spec to the standard — rewrites what was prohibited into the approved alternative, and adds what's required (e.g. the mandatory tests).
- Warns when it made that adjustment, so it's transparent what changed and why.
- Marks it
rejectedwhen the intent is irreconcilable with a guardrail (e.g. it explicitly asks for something prohibited). It's reviewable — you adjust the intent or the Owner revisits the guardrail.
That way governance happens at the source of the work, not as a veto at the end.