Salesforce

Build a single-user personal CRM that runs locally. No accounts, no cloud, no telemetry.

Stack, non-negotiable:
- Next.js (App Router) with TypeScript
- SQLite via better-sqlite3, file at ./data/crm.db
- Drizzle ORM for schema and migrations
- Tailwind CSS, no component library
- Runs with `npm run dev` on port 3000 and nothing else needs to be installed

Data model:
- companies: id, name, domain, notes, createdAt
- contacts: id, companyId (nullable), name, email, phone, role, notes, createdAt
- deals: id, title, companyId, contactId (nullable), amountCents, currency (default USD), stage, expectedCloseDate, createdAt, closedAt, outcome (won|lost|null)
- activities: id, kind (call|email|meeting|note), subject, body, occurredAt, contactId, dealId, createdAt
- tasks: id, title, dueDate, done, contactId, dealId
- Stages are a fixed list in code: lead, qualified, proposal, negotiation, closed.

Screens:
1. /pipeline: kanban board, one column per stage, deal cards showing title, company, amount. Drag between columns persists the stage. Column headers show count and summed amount.
2. /contacts and /companies: searchable tables, click through to a detail page showing linked deals, activity timeline newest first, and open tasks.
3. /deals/[id]: edit fields inline, log an activity, mark won or lost.
4. /today: tasks due today or overdue, plus deals with no activity in 14 days, labelled "going cold".

Also build:
- One global search box (cmd+K) over contacts, companies and deals, using SQLite LIKE queries. No vector search, no AI.
- A GET /api/export route that dumps everything as JSON, and a CSV import for contacts that maps columns in the UI before committing.
- Seed script with about 20 fake contacts, 8 companies and 12 deals so the board is not empty on first run.

Explicitly out of scope: email sending or syncing, OAuth, multi-user or permissions, workflow automation, reporting beyond the pipeline totals, mobile app, any third-party API. If you want any secret later, read it from .env and commit .env.example only.

Write a README with setup in under five commands and a one-paragraph note on how to back up the SQLite file.

# AGENTS.md — Build guide for Salesforce

## Project scope
A local web app with contacts, companies, deals on a drag-across pipeline, activity notes, and next-action reminders, backed by SQLite.

Catalogue verdict: no. You can absolutely one-shot a contacts-and-deals tracker, and for one person that is often better than Salesforce. But nobody pays Salesforce for a pipeline table. They pay for the thing everything else in the company already points at: the ETL jobs, the marketing automation, the support queue, the finance reports, the twelve years of closed-won history, the admin who wrote 400 validation rules, and the auditors who accept it. A personal replacement replaces the part of Salesforce that was never the expensive part. If you are a solo operator or a two-person shop, build it; if you are trying to leave an existing org, this is a migration project, not a coding session.
Use the implementation prompt below to define the deliverable. Complete each phase's acceptance checks before extending the scope.

## Working agreement
- Inspect the repository and its existing instructions before choosing paths, dependencies or commands. Keep one coherent stack and explain changes to the proposed architecture.
- Plan a vertical slice that accepts a real input and produces the useful output described below. Persist only the state the prompt calls for; respect memory-only and upstream-managed workflows. Use fixtures only when they are clearly labelled.
- After scaffolding, document the actual install, development, check and build commands in README and keep them synchronized with the package or project manifest. Do not report commands as successful unless they ran.
- Work in small steps. At handoff, list implemented flows, checks actually performed, remaining blockers, and any credentials or provider setup the owner must supply.
- Do not publish, spend money, contact customers, delete source data or run irreversible migrations without the project owner's authorization.

## Prerequisites
- Node runtime and a place to run it, local or a small VPS
- Optional: an IMAP or Gmail API credential if you want email logging, which is where the weekend goes

## Stack and architecture
- SQLite via better-sqlite3, file at ./data/crm.db.
- Data model:.
- Interface for this enterprise CRM platform workflow: a focused enterprise CRM platform input, review, and export interface.

## Security and data integrity
- Explicitly out of scope: email sending or syncing, OAuth, multi-user or permissions, workflow automation, reporting beyond the pipeline totals, mobile app, any third-party API. If you want any secret later, read it from .env and commit .env.example only.
- Validate imported records and file paths; keep user data within the storage location named in the prompt.
- State the practical limit: Every integration: marketing automation, support cases, quoting, billing, BI tooling, and the hundred connectors vendors ship for Salesforce and nothing else.

## Agent implementation rules
- Project rule, data: Data model:.
- Project rule, behavior: Runs with `npm run dev` on port 3000 and nothing else needs to be installed.
- Project rule, recovery: Explicitly out of scope: email sending or syncing, OAuth, multi-user or permissions, workflow automation, reporting beyond the pipeline totals, mobile app, any third-party API. If you want any secret later, read it from .env and commit .env.example only.

## Optional agent skills and references
No installable skill is listed for this guide; use the project rules and the stack documentation.

Read the linked SKILL.md and its dependencies before adding a skill. Select only the skills matching this project's runtime and task; their documentation does not supply API access, credentials or approval to perform external actions. Pin the reviewed revision where the tool supports it. Follow the chosen agent's documented project-level installation mechanism.

## Distribution ideas
These are optional planning notes. Obtain the owner's approval before publishing or contacting anyone.
- Demonstrate this enterprise CRM platform operation with sample or user-owned data: Runs with `npm run dev` on port 3000 and nothing else needs to be installed.
- Show a concrete result or export: A GET /api/export route that dumps everything as JSON, and a CSV import for contacts that maps columns in the UI before committing.
- State the practical limit: Every integration: marketing automation, support cases, quoting, billing, BI tooling, and the hundred connectors vendors ship for Salesforce and nothing else.

## Engineering roadmap
1. Phase 1, architecture and data: SQLite via better-sqlite3, file at ./data/crm.db. Data model:.
2. Phase 2, implement: Runs with `npm run dev` on port 3000 and nothing else needs to be installed.
3. Phase 3, implement: companies: id, name, domain, notes, createdAt.
4. Phase 4, review and output: contacts: id, companyId (nullable), name, email, phone, role, notes, createdAt. A GET /api/export route that dumps everything as JSON, and a CSV import for contacts that maps columns in the UI before committing.
5. Phase 5, recovery and acceptance: Explicitly out of scope: email sending or syncing, OAuth, multi-user or permissions, workflow automation, reporting beyond the pipeline totals, mobile app, any third-party API. If you want any secret later, read it from .env and commit .env.example only. Verify this invariant with a saved fixture: An invalid input or interrupted operation must retain the source and show a recoverable state; exported records must reload with the same IDs. State the practical limit: Every integration: marketing automation, support cases, quoting, billing, BI tooling, and the hundred connectors vendors ship for Salesforce and nothing else.

## Paid-product capabilities outside this build
- Every integration: marketing automation, support cases, quoting, billing, BI tooling, and the hundred connectors vendors ship for Salesforce and nothing else
- Multi-user permissions, sharing rules, field-level security, and audit trails that survive a compliance review
- Reporting, forecasting and dashboards that leadership already knows how to read
- The admin ecosystem: consultants, AppExchange, and a hiring pool that already knows the object model
- Historical data with lineage, which is the actual asset in any CRM older than a year

## Implementation prompt
Build a single-user personal CRM that runs locally. No accounts, no cloud, no telemetry.

Stack, non-negotiable:
- Next.js (App Router) with TypeScript
- SQLite via better-sqlite3, file at ./data/crm.db
- Drizzle ORM for schema and migrations
- Tailwind CSS, no component library
- Runs with `npm run dev` on port 3000 and nothing else needs to be installed

Data model:
- companies: id, name, domain, notes, createdAt
- contacts: id, companyId (nullable), name, email, phone, role, notes, createdAt
- deals: id, title, companyId, contactId (nullable), amountCents, currency (default USD), stage, expectedCloseDate, createdAt, closedAt, outcome (won|lost|null)
- activities: id, kind (call|email|meeting|note), subject, body, occurredAt, contactId, dealId, createdAt
- tasks: id, title, dueDate, done, contactId, dealId
- Stages are a fixed list in code: lead, qualified, proposal, negotiation, closed.

Screens:
1. /pipeline: kanban board, one column per stage, deal cards showing title, company, amount. Drag between columns persists the stage. Column headers show count and summed amount.
2. /contacts and /companies: searchable tables, click through to a detail page showing linked deals, activity timeline newest first, and open tasks.
3. /deals/[id]: edit fields inline, log an activity, mark won or lost.
4. /today: tasks due today or overdue, plus deals with no activity in 14 days, labelled "going cold".

Also build:
- One global search box (cmd+K) over contacts, companies and deals, using SQLite LIKE queries. No vector search, no AI.
- A GET /api/export route that dumps everything as JSON, and a CSV import for contacts that maps columns in the UI before committing.
- Seed script with about 20 fake contacts, 8 companies and 12 deals so the board is not empty on first run.

Explicitly out of scope: email sending or syncing, OAuth, multi-user or permissions, workflow automation, reporting beyond the pipeline totals, mobile app, any third-party API. If you want any secret later, read it from .env and commit .env.example only.

Write a README with setup in under five commands and a one-paragraph note on how to back up the SQLite file.

## Completion evidence
Demonstrate the prompt's acceptance scenarios against the scoped workflow. Include setup from a clean checkout and failure recovery. Check persistence across restart and export/restore only for the state the prompt says to store; for memory-only tools, confirm that temporary content is discarded as specified. Record actual results and remaining limitations. A detailed plan alone does not establish a working replacement.