Laravel crm ai skill

How to use the laravel-crm-ai-skill skill

Practical guide for using this skill. It's a complete reverse-engineering of a reference enterprise CRM, turned into an implementable specification for a Laravel CRM (built for a generic Laravel project, reusable in any app). No asset, branding, or proprietary data from the reference org: only patterns, UX, and logic were captured.

---

1. What it is (and isn't)

• It is: a product spec + UX + database + business logic + components + tests + roadmap, detailed enough that an agent can build the V1 of the CRM without reinventing the key decisions.
• It is NOT: already-implemented CRM code. Implementation happens by following the skill (it hasn't been written into an application repo yet).

2. Activation in Claude Code

• Automatic: working inside your Laravel project, the skill triggers on requests like "build the CRM leads view", "enterprise-style lead list", "opportunity detail page", "pipeline/kanban", "lead conversion", etc. (see the description field in SKILL.md).
• Manual: open SKILL.md (entry point) → follow the pointers into references/.

3. Where to start (reading order)

1. SKILL.md — overview, guiding principle, reuse map, guardrails, build order.
2. references/README.md — full file index + method + evidence legend.
3. references/definition-of-done.md — honest current state/percentages + what's still "to verify".
4. Then whichever domain file you need (see §5).

4. Evidence legend (used in every file)

Every important claim is labeled: Observed (seen in the UI) · Dynamically tested (by creating/editing demo records) · Deduced · Proposed for Laravel · To verify · Do not replicate · Implementation risk. → Trust "Tested/Observed"; treat "Proposed for Laravel" as a recommendation; verify "To verify" items before assuming them.

5. Usage paths

A) Understand the reference CRM you're replicating

source-crm-reference.md · screen-anatomy.md · objects-catalog.md · the various *-system.md files (list-view, filters, record-page, kanban, activities, conversion…).

B) Implement V1 in Laravel (main path)

1. laravel-implementation-blueprint.md — architecture, layers, action/policy/observer, naming, risks.
2. data-model-and-migrations.md — crm_* tables, fields, indexes, relations (additive, non-destructive).
3. component-catalog.md + ux-design-system.md + screen-anatomy.md — enterprise-style Blade/Alpine components + tokens.
4. modules.md — the concrete Lead module (routes/controller/views/quick-actions) that every other object mirrors.
5. permissions-and-roles.md + testing-strategy.md — policy + tests.
6. v1-v2-v3-roadmap.md — what to build in V1 (NOT everything!) and what to defer.
7. Build order (in SKILL.md): Phase A (Lead layer: extract CommercialPipelineActionService + additive migration + /crm/leads) → Phase B (Account/Contact/Opportunity + conversion + activities).

C) Build a single module/screen

Open the specific file: list-view-system.md, record-page-system.md (+ lead/account/contact/opportunity-record-page.md), filter-builder-system.md, kanban-system.md, stage-transition-system.md, lead-conversion-system.md, activity-and-timeline-system.md, inline-edit-system.md, form-and-modal-system.md, validation-and-error-system.md, field-catalog.md.

D) Complete remaining tests against a reference demo org (optional)

See definition-of-done.md §remaining + open-questions-and-assumptions.md. Recreate a small demo dataset with the CRM-SKILL- prefix and fake data only (e.g. @crm-skill.test emails), then clean it up at the end (procedure in demo-data-register.md). Never use real accounts, real credentials, or real personal data for this — see the guardrails in SKILL.md.

6. Golden rules (from the guardrails)

• Do NOT clone the reference enterprise platform itself (Flow, report builder, AI, complex sharing) → v1-v2-v3-roadmap.md §Do not replicate.
• Reuse commercial_pipeline_entries + the existing next-action engine; add crm_* tables additively and reversibly; don't break the existing pipeline/today/import-export.
• Server-side first (filters/sort/pagination in SQL, no N+1); Alpine only for interactions.
• Policy on every route/action; server-side validation; no real reference-CRM data in the repo/commits/seeds.
• Fixed stack: Laravel 13 · Blade · Alpine · Tailwind · SQLite. New dependencies only in V2 (SortableJS for Kanban, FullCalendar for events).

7. Quick file map (by need)

You need…	Open
Index + method	references/README.md
Status/progress	references/definition-of-done.md
The real reference anatomy	references/source-crm-reference.md, screen-anatomy.md
Objects + relationships	references/objects-catalog.md, data-model-and-migrations.md
UI components	references/component-catalog.md, ux-design-system.md
Lists/filters/kanban	references/list-view-system.md, filter-builder-system.md, kanban-system.md
Record/detail	references/record-page-system.md (+ per-object), inline-edit-system.md
Flows	references/lead-conversion-system.md, stage-transition-system.md, activity-and-timeline-system.md
Implementation	references/laravel-implementation-blueprint.md, modules.md
Permissions/testing	references/permissions-and-roles.md, testing-strategy.md
Priorities	references/v1-v2-v3-roadmap.md

8. Suggested next step

Ask your AI assistant, inside your Laravel project: "implement Phase A of the CRM following the laravel-crm-ai-skill skill" → it will start with extracting the service + an additive migration + /crm/leads, in TDD, without touching existing functionality.