Configured for AI Agents
FRAME is designed not only for a strong developer experience, but to be highly understandable and predictable for Artificial Intelligence (AI) assistants and generative coding tools.
Clear Documentation and Rules
Section titled “Clear Documentation and Rules”In an environment where AI assists development, context is everything. That is why FRAME provides:
AGENTS.mdfiles: Hierarchical context at the root and in each app/package (CLAUDE.md→AGENTS.mdsymlink).- Skills under
.agents/skills/: actionable domain guides with metadata that lets agents discover and load relevant instructions on demand. - Predictable Structure: packages and apps follow stable export patterns and layout.
- Safety rails: Strict TypeScript + Oxlint + Lefthook correct deviations before merge.
Review-Centered Workflows
Section titled “Review-Centered Workflows”FRAME promotes a collaborative model where AI and developers have complementary roles. Rather than aiming for full automation, the ecosystem treats generative coding as an acceleration engine that proposes solutions, structures components, and performs refactors.
Under this approach, technical review is the central step of development. All of FRAME’s infrastructure — from strict Drizzle and Zod typing to Oxlint’s millisecond linting — is specifically designed to speed up and strengthen this validation stage. AI drafts and proposes, allowing the developer to focus energy on auditing logic, evaluating security, and making architectural decisions.
Verify the agent environment
Section titled “Verify the agent environment”From the agent chat: invoke the ai-setup skill (“test AI setup”) for a briefing on context and skills.