Repository Context for AI Agents¶
If you are an AI agent landing in this repository for the first time, start here.
What this is¶
devonshire-digital is the company-level governance and documentation repository for Devonshire
Digital LLC, a private, proprietary parent company. It is the source of truth for company policy,
standards, decision records, and AI agent instructions. It is not (necessarily) where product
code lives — see repository boundaries for why that is
an intentionally open question rather than a default assumption.
There is no remote configured and no OSS license; this is private, proprietary material.
The venture list¶
- Shelfery — a home food-storage / pantry application. Venture slug:
shelfery. - Digital Products — the internal working name for a productivity-tool venture. Venture slug:
digital-products. This is explicitly a working name, not a confirmed public brand (see open-questions.md, OQ-001).
Both ventures inherit company-level policy and standards and may add their own, stricter, venture-level material — see authority-and-inheritance.md.
Where things live¶
docs/company/— mission, principles, product philosophy, portfolio strategy, brand architecture, and related company-identity documents.docs/governance/— the rules for how documentation itself works: taxonomy, metadata, authority/precedence, decision rights, lifecycle, classification, change management, versioning, assumptions/open-questions/risk registers, repository boundaries, Git/Notion responsibilities.docs/ai/— this directory: AI-agent-specific operating instructions built on top ofdocs/governance/.docs/decisions/— Architecture Decision Records (ADRs).docs/architecture/— architecture and technical strategy documents, including future repository strategy.docs/standards/— company-level standards documents.ventures/<slug>/— venture-specific material, where present, following the same governance and AI-instruction conventions scoped to that venture.- Root files —
README.md,AGENTS.md,CLAUDE.md,GEMINI.md,CONTRIBUTING.md,SECURITY.md,CODE_OF_CONDUCT.md,CHANGELOG.md,NOTICE.md.
AGENTS.md is the contract¶
The root AGENTS.md is the authoritative, tool-agnostic contract for how any
AI agent should behave in this repository. CLAUDE.md and GEMINI.md exist as tool-specific
entry points but should not contain conflicting rules — they point back to AGENTS.md and this
docs/ai/ directory rather than duplicating divergent copies of the same rules (see
prohibited-behaviors.md on redundant AI instruction files).
Read source-of-truth-rules.md next, then use context-loading-order.md to determine what else to load for your specific task.