Available Context

@_platform-references/org-variables.md

Ops Table Manager

Goal

Manage the lifecycle and structure of Ops tables -- the flexible, spreadsheet-like data stores that power prospecting workflows. This skill handles creating tables, listing existing tables, viewing table details and columns, adding columns, and deleting tables.

Ops tables are the foundation of the prospecting system. Every enrichment, AI query, and data import flows through them. Getting the structure right upfront saves reps from painful restructuring later.

Required Capabilities

• Ops Tables: CRUD access to ops tables and column management

Inputs

• table_id: ID of an existing table (for view, modify, or delete operations)
• table_name: Name for a new table (for create operations)
• column_name: Name for a new column (for add column operations)
• column_type: Type of column -- one of: text, number, url, email, date, select, ai_generate
• description: Optional description for a new table
• columns: Optional array of initial columns when creating a table (each with name and type)

Instructions

Listing Tables

When the user wants to see their tables:

1. Call execute_action("list_ops_tables", { limit: 50 })
2. Present the results as a scannable list showing:
   • Table name
   • Row count
   • Source type (manual, import, apollo, hubspot, etc.)
3. If no tables exist, suggest creating one: "You don't have any ops tables yet. Want me to create one?"

Viewing Table Details

When the user asks about a specific table:

1. Identify the table -- if the user gives a name, first call list_ops_tables to find the matching ID
2. Call execute_action("get_ops_table", { table_id: "<id>" })
3. Present:
   • Table name and description
   • Column list with types
   • Row count
   • Enrichment stats (if any columns have been enriched)

Creating a Table

When the user wants a new table:

1. Confirm the table name. If not provided, ask: "What would you like to name the table?"
2. Ask about initial columns if not specified. Suggest sensible defaults based on context:
   • For a leads table: Name, Company, Title, Email, LinkedIn URL, Phone
   • For a companies table: Company Name, Domain, Industry, Size, Location
3. Call execute_action("create_ops_table", { name: "<name>", description: "<desc>", columns: [...] })
4. Confirm creation with the table name and column count

Adding a Column

When the user wants to add a column:

1. Identify the target table (by name or ID)
2. Determine column name and type. If type is not specified, infer from the name:
   • "email" -> email
   • "website", "url", "linkedin" -> url
   • "revenue", "count", "size", "score" -> number
   • "date", "founded", "created" -> date
   • Otherwise -> text
3. Call execute_action("add_ops_column", { table_id: "<id>", name: "<name>", column_type: "<type>" })
4. For ai_generate type columns, ask for the generation prompt: "What should the AI generate for this column?"
5. Confirm the column was added

Deleting a Table

When the user wants to delete a table:

1. Always confirm before deleting. Show the table name and row count: "This will permanently delete '[Table Name]' with X rows. Are you sure?"
2. Only proceed after explicit confirmation
3. Call execute_action("delete_ops_table", { table_id: "<id>", confirm: true })
4. Confirm deletion

Available Actions

Action	Parameters	Returns
list_ops_tables	{ limit?: number }	Array of tables with id, name, row_count, source_type
get_ops_table	{ table_id: string }	Table with columns array and enrichment stats
create_ops_table	{ name: string, description?: string, columns?: Array<{name, type}> }	New table object
delete_ops_table	{ table_id: string, confirm: true }	Deletion confirmation
add_ops_column	{ table_id: string, name: string, column_type: string, config?: object }	New column object

Output Format

List Tables Response

YOUR OPS TABLES (X tables)
  Lead Prospects    142 rows   Source: apollo
  Target Companies   38 rows   Source: manual
  HubSpot Import     95 rows   Source: hubspot

Table Details Response

TABLE: Lead Prospects (142 rows)
Source: apollo | Created: Jan 15, 2026

COLUMNS (8)
  Name         text
  Company      text
  Title        text
  Email        email
  LinkedIn     url
  Phone        text
  Score        number
  Notes        ai_generate (enriched: 89%)

Create Confirmation

Table "[Name]" created with X columns.
Ready to add data -- you can import from Apollo, HubSpot, or add rows manually.

Error Handling

Table not found

If a table name doesn't match any existing table, list available tables and ask: "I couldn't find a table called '[name]'. Here are your tables -- which one did you mean?"

Duplicate table name

If the user tries to create a table with a name that already exists, inform them and suggest alternatives: "A table called '[name]' already exists. Want me to use a different name, or did you mean to open the existing one?"

Delete confirmation not given

If the user says "delete table" but does not explicitly confirm after the warning, do NOT proceed. Respond: "No problem, the table is safe. Let me know if you change your mind."

Invalid column type

If an unrecognized column type is requested, suggest the closest valid type and confirm before proceeding.

Guidelines

• Always use table IDs (not names) when calling actions -- names are for display only
• When creating tables for prospecting, suggest the standard lead columns (Name, Company, Title, Email, LinkedIn)
• Column names should be human-readable (use "Company Name" not "company_name")
• Suggest ai_generate column type when the user wants AI-derived data (e.g., "company summary", "personalized intro")
• Never delete a table without explicit user confirmation