Prompts

Prompts are pre-built templates that structure how an AI assistant analyzes your CRM data. Instead of crafting the perfect question each time, invoke a prompt and the assistant is handed a step-by-step plan that pulls the right resources and calls the right tools automatically.

The Kasar MCP server exposes exactly three prompts: contact_summary, pipeline_review, and weekly_report. Each one returns a single user message that instructs the assistant which resources to read and which tools to call, then how to synthesize the result.

contact_summary

Generate a comprehensive summary of a contact: recent interactions, open tasks, notes, and associated deals.

Arguments

Invocation

Ask your AI assistant:

"Use the contact_summary prompt for contact 9f1c2e7a-..."

Or in an MCP client that supports prompt selection, choose Contact Summary and provide the contact UUID.

What the prompt does

The generated plan instructs the assistant to:

1. Read kasar://schema/contacts to learn the available fields.
2. Call crm_find_records with object_name: "contacts" and record_id set to the contact UUID to fetch the contact details.
3. Call crm_find_activity with object_name: "contacts" and record_id set to the contact UUID to pull recent interactions (emails, LinkedIn messages, calls), open and overdue tasks, and team notes.
4. Check whether the contact is linked to any deals/opportunities.

It then synthesizes a structured summary:

- **Contact Overview**: Name, company, role, contact info
- **Recent Activity**: Last 5 interactions with channel and date
- **Open Tasks**: Any pending or overdue tasks
- **Key Notes**: Important notes from the team
- **Deals**: Associated opportunities and their pipeline stage
- **Suggested Next Actions**: Based on the activity pattern

pipeline_review

Analyze a sales pipeline: deals by stage, stagnant deals, conversion rates, and revenue forecasts.

Arguments

Invocation

Ask your AI assistant:

"Run a pipeline review"

Or specify a pipeline:

"Use the pipeline_review prompt for the Enterprise pipeline"

What the prompt does

The generated plan instructs the assistant to:

1. Read kasar://pipelines to get the pipeline stages.
2. Call crm_aggregate_records with object_name: "opportunities" grouped by pipeline step to get the count of deals per stage and the total value per stage (sum on the amount field).
3. Call crm_find_records to find stagnant deals (opportunities not updated in over 14 days, still in open stages).
4. Call crm_find_records to retrieve recently won and lost deals.

It then produces a pipeline review:

- **Pipeline Health**: Total deals, total value, average deal size
- **Stage Distribution**: Deals and value per stage (table format)
- **Conversion Funnel**: Estimated conversion rates between stages
- **Stagnant Deals**: Deals stuck > 14 days with suggested actions
- **Recent Wins/Losses**: Deals closed this month
- **Revenue Forecast**: Expected close based on stage probabilities
- **Recommendations**: Top 3 actions to improve pipeline flow

weekly_report

Generate a weekly activity report: new contacts, deal progress, completed tasks, and key interactions.

Arguments

Invocation

Ask your AI assistant:

"Generate a weekly report"

Or for a specific week:

"Use the weekly_report prompt for 2026-W14"

What the prompt does

The generated plan instructs the assistant to:

1. Call crm_find_records on contacts filtered by created_at >= start of week to count new contacts.
2. Call crm_find_records on companies filtered by created_at >= start of week to count new companies.
3. Call crm_aggregate_records on opportunities to get deals created, won, and lost this week.
4. Call crm_task_stats to get task completion metrics.
5. Call crm_manage_tasks with action: "list" and mode: "completed" to see what was done.

It then compiles a weekly report:

- **Summary**: One-paragraph executive overview
- **New Contacts & Companies**: Count and notable additions
- **Pipeline Activity**: Deals created, advanced, won, lost with values
- **Task Completion**: Tasks completed vs created, completion rate
- **Key Interactions**: Most active contacts/threads this week
- **Next Week Focus**: Suggested priorities based on this week's activity