Start by entity
Use the index when the question is “what is this thing and why does it matter?” rather than “which Rust crate owns it?”
Why entity-first lowers labor cost
Labor cost falls when every workflow starts from named, traceable operating facts
Across a 170-site pet-resort operating context, repeated manager and front-desk work appears in morning dashboard reconciliation, checkout exception review, retention prioritization, source-data cleanup, and repetitive customer or staff drafts. The entity atlas turns those repeated lookups into named things a reviewer can understand: source systems, provenance, pets, reservations, service-line contracts, care facts, review gates, workflow packets, blocked actions, outcome records, and labor-minute evidence.
The useful AI pattern is not a free-form chatbot and not crate-first architecture. It is source-grounded workflow automation with review gates and measured outcomes. Agents only summarize, rank, draft, and recommend from app-owned packets; the app/domain contracts and human/system-of-record approvals decide what can happen.
Front-door navigation
Choose an entity, role, workflow, safety boundary, or evidence path
How a non-coder should enter
The repository models operating facts, source evidence, review gates, draft workflows, storage records, and integration boundaries so AI agents can help staff without owning business truth. Start from the business thing you recognize, then use code-derived evidence only after the operating meaning is clear.
- Entity index: choose the important thing and see meaning, authority, automation, blocked actions, value, and proof.
- Audience paths: operations, resort manager, IT/integration, safety/compliance, and product/customer-success routes.
- Workflow-to-entity navigation: enter from labor loops and return to the entities each loop depends on.
- Safety boundaries: read what agents may draft/recommend and what must remain human-reviewed or blocked.
- Evidence links: inspect source/Rustdoc/test proof without making crate names the front door.
Audience paths
Start with the path that matches the decision you need to make, then use the entity index and Rustdocs as code-derived evidence after the business explanation.
- Operations leader: labor saved, status, and human-owned approvals.
- AI program evaluator: governed agent/app workflow pattern.
- Safety/compliance reviewer: blocked side effects, review gates, and audit evidence.
- Gingr/source-data owner: source-system evidence, provider mappings, and no blind PMS mutation.
- Developer: code-derived contract pages and implementation paths.
Entity-first operating model
The entity atlas is the public docs spine
The canonical entity index is the shared spine for README and public-doc navigation. It lets a non-coder choose a source fact, customer, pet, reservation, service-line contract, review gate, workflow packet, outcome record, Gingr provider boundary, or runtime/storage surface and see what it is, why it saves labor, who is authoritative, what an agent may draft, what remains blocked, how value is measured, and where source/Rustdoc/test proof lives.
Follow relationships
Trace the safe flow from provider/staff evidence to provenance, domain truth, app packet, agent draft, human review, outcome record, and storage/API proof.
Use source-backed families
Family pages keep plain-English operations, blocked actions, value metrics, and source/Rustdoc evidence together without making Markdown the behavior authority.
Crosswalk from entity to proof
Use the contract crosswalk when the question is “show me where this entity enters, normalizes, appears in workflows, persists, and is exposed.”
Evidence anchors for page writers
When adding or checking entity pages, use the evidence anchor map to keep public claims tied to source, generated Rustdocs, tests, and explicit gaps.
Glossary translation layer
Use nearby term explanations before reading Rust names
When a non-coder sees repo terms such as domain, workflow packet, source ref, DTO, or blocked action, these glossary links translate the code-derived contract into pet-resort operational meaning without turning it into a detached dictionary.
Architecture layers
domain is business truth, app builds workflow/review bundles, and storage persists projections and outcomes.
Source evidence
Gingr, provider records, DTOs, source refs, and provenance explain what evidence exists and what it still does not authorize.
Workflow safety
workflow packets, drafts, review gates, blocked actions, and outcome capture explain why agent help stays reviewable and measured.
- Glossary translation summary for readers moving between landing pages, operator workflow docs, and generated Rustdocs.
- Core entity-linked term explanations for cross-cutting terms that touch authority, evidence, safety, automation, value measurement, and Rustdoc proof.
- Glossary-to-entity link convention for downstream docs workers who need the current cross-linking pattern.
Workflow-to-entity map
Workflows are entry paths back to entities
For non-coders who arrive through a labor loop, use the operator workflow index and workflow-to-entity summary before Rustdoc. They link Manager Daily Brief, Booking Triage, Data Quality Hygiene, Checkout Completion, Grooming Rebooking / Retention, Daily Updates / Pawgress Drafts, and Regional Labor Exceptions back to the featured entities, related entities, contracts, authority, approvals, blocked actions, outcomes, and source/Rustdoc/test backing each workflow depends on.
Manager Daily Brief
A GM/AGM starts the day with ranked actions for demand-versus-staffing, checkout exceptions, retention follow-up, and data-quality issues. Each action carries source evidence, owner persona, review gate, blocked side effects, and estimated/actual labor-minute tracking.
Data-Quality Hygiene
Missing or ambiguous source facts become visible cleanup work instead of repeated downstream rework. The workflow ranks internal hygiene tasks, preserves ambiguity for review, rejects customer/provider side effects, and measures manual reconciliation minutes avoided.
Retention, grooming, and daily-update drafts
The app can identify source-grounded, consent-aware opportunities and let agents draft staff-reviewed follow-up or customer-update language. It does not authorize live customer sends, discounts, payment movement, scheduling, or provider updates without explicit app-owned approval.
Gingr/source-system integration
Gingr is treated as source evidence, not unchecked domain truth. Provider ids, endpoint shapes, webhooks, DTOs, mappings, redaction, and documented gaps are kept at the integration boundary before app workflows consume normalized facts.
Service-line and support summaries
Boarding, Daycare, Grooming, Training, Retail, Reservations/checkout, and care/document/temperament/incident support each have concise README or Rustdoc entrypoints so non-coders can find the staff queue, labor-saving claim, source authority, and review gates without reading every type first.
- Boarding, Daycare, Grooming, Training, and Retail Rustdocs.
- Reservations, booking triage, and checkout completion Rustdocs.
- Care, documents, vaccines, temperament, and incidents Rustdocs.
Ready vs planned
Current status without overclaiming
| Stage | What this repo can say | Evidence |
|---|---|---|
| local/demo ready | Source-grounded context to reviewable draft to deterministic validation to reviewed outcome capture can be demonstrated with fake/local data. | Manager Daily Brief Rustdocs, Data-quality Rustdocs, local smoke docs. |
| sandbox continuation | The contract pattern is ready for more internal, read-only, fixture-safe, or sandbox workflows. | Agent/app infrastructure guide and the labor crosswalk. |
| pilot work remains | Pilot work still needs real source/read-model access, identity and approval records, durable audit retention, monitoring, rollback, operator metric definitions, and staff review UI. | Labor-cost readiness memo. |
| live/member-facing blocked | No live customer sends, provider/PMS writes, schedule changes, payment/refund/discount movement, hidden ambiguity resolution, or safety-sensitive decisions are authorized by current repo contracts. | Safety boundary, policy Rustdocs, workflow Rustdocs. |
Choose your path
Audience paths through the entity atlas
Pick the path that matches the decision you need to make, then use the entity atlas and Rustdocs as proof. The full path guide explains which entities to read first, what automation may draft or recommend, what remains human-reviewed, and how labor-cost value is measured.
Operations leaders
Decide which labor loops to prioritize across resorts: manager brief, data-quality hygiene, retention, checkout exceptions, or draft-writing reduction.
- Read first: operations portfolio, manager brief outcome records, data-quality issues, review gates, storage outcomes.
- Human-reviewed: staffing, money, schedules, customer/provider side effects, safety exceptions.
- Value proof: actual minutes, reporting group, completed/deferred/suppressed/wrong-source outcomes.
Resort managers
Route daily queue work to front desk, manager, care reviewer, approved sender, or payment/source owner without losing the safety boundary.
- Read first: reservations, customers, pets, care/vaccine/document facts, booking/checkout/daily-update/retention packets.
- Human-reviewed: booking/status changes, customer sends, care/safety approvals, payments, policy exceptions.
- Value proof: fewer source lookups, faster handoffs, draft-writing time, reviewed dispositions.
IT and integration readers / Gingr/integration owner
Judge how source systems such as Gingr are modeled as evidence without pretending provider records are canonical truth or mutating live systems.
- Read first: source refs, provenance, Gingr endpoints/DTOs/mappings/webhooks, data-quality issues, runtime/storage surfaces.
- Human-reviewed: provider/PMS writes, credential exposure, hidden source cleanup, unsupported endpoint assumptions.
- Value proof: reconciliation minutes, hygiene outcomes, wrong-source counts, fewer integration handoffs.
Compliance and safety reviewers / safety/compliance reviewer
Verify what agents cannot do, who approves risky actions, and where source ambiguity, audit evidence, and blocked actions are preserved.
- Read first: review gates, blocked actions, source/provenance, care/vaccine/incident/message facts, agent specs, tool ports.
- Human-reviewed: all live customer, provider, money, schedule, safety, medical, incident, and policy-exception side effects.
- Value proof: reviewed outcomes, audit records, source refs, omitted facts, blocked-action reasons.
Product and customer success / AI program evaluator
Tell honest product stories around reviewable queues and drafts while defining the proof a pilot or customer-success plan must capture.
- Read first: operations portfolio, workflow packets, customer/pet/reservation/message entities, revenue opportunities, outcome records.
- Human-reviewed: outreach, offers/discounts, appointments, payments, service promises, safety conclusions, live labor-saving claims.
- Value proof: role minutes, avoided rework, conversion/suppression outcomes, customer-message dispositions.
Safety boundary
Rustdocs are evidence, not production authorization
Agents are review-gated and do not perform live member, customer, provider/PMS, schedule, payment, refund, discount, incident, vaccine, medical, safety, or personnel side effects without deterministic app-owned approval contracts and human review. Source ambiguity is preserved for review; it is never hidden by agent prose.
Rustdocs as code-derived evidence
Inspect the compiled contracts behind the narrative
| Operational translation | Technical surface |
|---|---|
| Pet-resort facts, policies, review gates, source evidence, and labor/safety concepts staff can trust. | domain crate |
| Manager/front-desk workflows, review packets, draft validation, and agent-safe recommendations. | app crate |
| Outcome records, audit evidence, normalized projections, and explicit storage-to-business conversions. | storage crate |
| Gingr/source-system requests, provider payloads, webhooks, mappings, and source-data boundaries. | gingr crate |
| Thin runtime shells for local demos, workers, and operator/developer inspection. | API shell, worker shell, CLI shell |
Source facts
Review gates
Manager brief
app::manager_daily_brief, domain::daily_brief, storage::operations