carta-compensation-scorecard

carta-cap-table:6.103.1

CTC Scorecard

Show how a corporation's compensation stacks up against market — at the corp level (band distribution rollup) or at the employee level (per-employee compa-ratio table) — using Carta Total Compensation data.

This skill calls compensation-service's scorecard endpoints (the same endpoints the CTC in-product UI uses) and renders the result in chat. By construction, the output matches what the customer sees in their CTC product surface.

CRITICAL — Casing rule for ALL user-facing CTC values.

Render CTC taxonomy values in Title Case display form, never the UPPER_SNAKE_CASE API enums. Matches the carta-compensation-rolematcher and carta-compensation-benchmarks convention so the plugin's voice is consistent across skills.

Field	Use in user-facing text	Never
Job area	the job's label from compensation:get:filter_options	the job's value
Focus	DevOps and Site Reliability, Account Executive, Financial Planning and Analysis — the exact API focus value, the same in text and in the API call	devops and site reliability, ACCOUNT_EXECUTIVE, fp&a — wrong in text and in the API call
Level	Title Case of the level's value (Mid 1, Senior 1, VP 1, C-Level)	the level's value from compensation:get:filter_options
Track	IC, Manager, Executive, Unknown	ic, manager, executive, UNKNOWN
Band	Low, Mid, High	low, mid, high

The value forms (UPPER_SNAKE_CASE) are only for machine handoff to MCP commands (the job_filters parameter). Switch to the labels before any value reaches the user. The leader parameter is a boolean.

Bands — a response score is LOW, MID or HIGH. A band says where an employee's compa-ratio sits against the market range for their role: LOW is below the range, MID is within it, HIGH is above it. It is not a percentile, so a percentile-specific ask ("below P50", "above P75") is not a band ask — see Percentile asks in Step 4b.

Response value	Display	Maps to user phrasing
LOW	Low	"below market"
MID	Mid	"at market"
HIGH	High	"above market"

Parse LOW / MID / HIGH from response score fields. The values to send as the score filter come from params.score[].value in Step 2b. Display the actual band (Low / Mid / High) in tables — do not relabel it to "Below market". When the user asked for a band by market phrasing (e.g. "show me below-market employees"), echo their phrasing in the surrounding prose AND map it through params.score, e.g. "Here are your below-market (Low) employees:" — but the per-row Band column still shows Low.

Use MCP, not CLI. Every API call in this skill goes through the carta MCP. Do not shell out to the carta CLI — that bypasses the formatters, the 403 handler, and the attribution requirement.

⚠️ Every compensation command goes through read_tool — copy the tool names exactly.

Examples below use the shorthand read_tool({...}) — read it as mcp__carta__read_tool({"name": ..., "arguments": {...}}). Do not use fetch or discover: both are deprecated and usually not registered, so they fail with "No such tool available".

The tool name is the command name with colons turned into __. Hyphens and underscores inside a segment stay as they are. The names are irregular, so never guess one:

Command	Tool name (pass as name)	✗ Not a real tool
compensation:get:subscription_status	compensation__get__subscription_status	—
compensation:get:filter_options	compensation__get__filter_options	—
compensation:get:plan	compensation__get__plan	—
compensation:get:employee-scorecard	compensation__get__employee-scorecard (hyphen)	compensation__get__employee_scorecard → "Unknown tool"
compensation:get:corporation-scorecard	compensation__get__corporation-scorecard (hyphen)	compensation__get__corporation_scorecard → "Unknown tool"
context_tools:get:profile	context_tools__get__profile	—

Arguments go in arguments (not params), and page_size is snake_case.