Available Context

@_platform-references/org-variables.md

Ops Data Manager

Goal

Work with the actual data inside Ops tables -- viewing rows, adding new leads or records, and updating individual cell values. This is the data layer that sits on top of table structure (managed by ops-table-manager).

Sales reps spend most of their time interacting with table data, not table structure. They need to quickly view their prospect lists, add new leads from conversations, and fix or update individual data points without leaving the chat.

Required Capabilities

• Ops Tables: Read/write access to ops table data

Inputs

• table_id: ID of the ops table to work with
• rows: Array of row objects for bulk adding (each row is a Record<string, string> mapping column names to values)
• row_id: ID of a specific row (for cell updates)
• column_id: ID of a specific column (for cell updates)
• value: New value for a cell update
• limit: Number of rows to fetch (default 25)
• offset: Pagination offset for viewing data
• filter: Filter criteria for viewing data

Instructions

Viewing Table Data

When the user wants to see data in a table:

1. Identify the target table. If the user says a table name, resolve it to an ID using list_ops_tables first.
2. Call execute_action("get_ops_table_data", { table_id: "<id>", limit: 25, offset: 0 })
3. Present data in a clean, tabular format showing the most important columns first
4. If there are more rows than the limit, note the total: "Showing 25 of 142 rows. Want me to show more?"
5. If the user asks for filtered data, use the filter parameter

Adding Rows

When the user wants to add data:

1. Identify the target table
2. Parse the data the user wants to add. Supported formats:
   • Explicit structured data: "Add John Smith, VP Sales at Acme Corp, john@acme.com"
   • Multiple leads: "Add these three leads: ..."
   • From conversation context: "Add the leads we just found"
3. Map the data to column names. Use get_ops_table to check the table's columns if needed.
4. Call execute_action("add_ops_rows", { table_id: "<id>", rows: [...] })
   • Each row is an object with column names as keys: { "Name": "John Smith", "Company": "Acme Corp", "Title": "VP Sales", "Email": "john@acme.com" }
5. Confirm: "Added X rows to [Table Name]."

Updating a Cell

When the user wants to change a specific value:

1. Identify the row and column. The user might say:
   • "Change John Smith's email to john.smith@newdomain.com"
   • "Update the title for row 3 to Director"
   • "Set the score to 85 for Acme Corp"
2. Resolve the row and column IDs. If the user references by name, fetch the table data to find the matching row_id and column_id.
3. Call execute_action("update_ops_cell", { row_id: "<row_id>", column_id: "<column_id>", value: "<new_value>" })
4. Confirm: "Updated [Column Name] for [Row Identifier] from '[old]' to '[new]'."

Bulk Operations

When adding multiple rows:

1. Parse all the data first before making any calls
2. Validate that the column names match the table structure
3. Add all rows in a single add_ops_rows call (the action handles batching internally)
4. Report total rows added and any that failed validation

Available Actions

Action	Parameters	Returns
get_ops_table_data	{ table_id: string, limit?: number, offset?: number, filter?: object }	Rows with column values and pagination
add_ops_rows	{ table_id: string, rows: Array<Record<string, string>> }	Count of rows added
update_ops_cell	{ row_id: string, column_id: string, value: string }	Updated cell confirmation

Output Format

View Data Response

LEAD PROSPECTS (showing 25 of 142)

Name              Company        Title           Email                  Score
John Smith        Acme Corp      VP Sales        john@acme.com          85
Sarah Chen        TechFlow       Director Eng    sarah@techflow.io      72
Mike Ross         DataBridge     CTO             mike@databridge.com    91
...

Showing rows 1-25. Say "show more" for the next page.

Add Rows Confirmation

Added 3 rows to "Lead Prospects":
  - John Smith (Acme Corp)
  - Sarah Chen (TechFlow)
  - Mike Ross (DataBridge)

Table now has 145 rows total.

Update Cell Confirmation

Updated Email for John Smith:
  Old: john@oldcompany.com
  New: john@acme.com

Error Handling

Table has no data

If the table exists but has no rows: "This table is empty. Want me to add some rows, or import data from Apollo/HubSpot?"

Column name mismatch

If the user provides data with column names that don't match the table: "The table doesn't have a '[column]' column. Available columns are: [list]. Did you mean '[closest match]'?"

Row not found

If the user references a row that doesn't exist: "I couldn't find a row matching '[identifier]'. Let me show you the current data so we can find the right one."

Invalid value for column type

If the user tries to set a non-numeric value in a number column or an invalid email in an email column, flag it: "That doesn't look like a valid [type] for the [column] column. Did you mean '[suggestion]'?"

Guidelines

• When displaying data, prioritize columns: Name/Company first, then email/contact info, then other fields
• For large tables (100+ rows), always paginate -- never dump the entire table
• When adding rows, validate against the table's column structure before calling the action
• Map user-friendly column references to actual column IDs (users say "email", not "col_abc123")
• If the user says "add a lead" (singular), still use add_ops_rows with a single-item array
• Preserve existing cell values when updating -- never clear other cells in the same row