[VENTURE_NAME] — AI Guardrails¶
Template placeholder. Company-level guardrails (root
docs/) always apply and are not repeated here in full — this document only adds venture- specific guardrails, if any are needed.
Things agents must not do in this venture without explicit human approval¶
- [Placeholder guardrail 1 — e.g. domain-specific risk unique to this venture]
- [Placeholder guardrail 2]
Things agents may do routinely without asking¶
- [Placeholder — routine, low-risk action 1]
- [Placeholder — routine, low-risk action 2]
Escalation¶
If a task in this venture appears to require an exception to a company-level
guardrail, do not proceed — escalate per the root AGENTS.md escalation
process instead of resolving it locally.
Relationship to company-level guardrails¶
This document supplements, and never overrides, company-level guardrails
defined in the repository root docs/. If no venture-specific guardrails
exist yet, the company-level guardrails alone govern this venture.