:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
import type { KPIData, ToolRow } from '../mcpDashboardOverviewLogic'
import { buildChips, buildEditorPrompt, buildHeadline, buildMaxPrompt } from './firstLookCopy'
const metric = (value: number): KPIData['sessions'] => ({
value,
previousValue: 0,
deltaPct: null,
sparkline: [],
sparklineLabels: [],
goodDirection: 'up',
})
const KPIS: KPIData = {
sessions: metric(120),
toolCalls: metric(1480),
errorRatePct: metric(9),
p95LatencyMs: metric(3525),
}
const tool = (over: Partial<ToolRow> = {}): ToolRow => ({
tool: 'execute-sql',
total_calls: 1480,
errors: 144,
error_rate_pct: 9.7,
p95_duration_ms: 3525,
...over,
})
describe('firstLookCopy', () => {
describe('buildHeadline', () => {
it('names the company, dominant client, busiest tool, and flags a flaky one', () => {
const headline = buildHeadline({
company: 'Acme',
project: 'prod',
topTool: tool(),
client: 'Claude Code',
kpis: KPIS,
})
expect(headline).toContain("Acme's MCP server is live")
expect(headline).toContain('Claude Code agents')
expect(headline).toContain('execute-sql')
expect(headline).toContain('errors 9.7% of the time')
})
it('omits the client phrase and the flaky clause when not warranted', () => {
const headline = buildHeadline({
company: null,
project: 'prod',
topTool: tool({ error_rate_pct: 1 }),
client: null,
kpis: KPIS,
})
expect(headline).toContain("prod's MCP server is live")
expect(headline).toContain('Agents leaned on')
expect(headline).not.toContain('agents leaned on') // no client prefix
expect(headline).not.toContain('errors')
})
it('falls back to top-line counts when no per-tool rows resolved', () => {
const headline = buildHeadline({ company: null, project: null, topTool: null, client: null, kpis: KPIS })
expect(headline).toContain('Your MCP server is live')
expect(headline).toContain('tool calls across')
expect(headline).toContain('sessions so far')
})
})
describe('buildMaxPrompt', () => {
it('references the worst error tool and its rate', () => {
const prompt = buildMaxPrompt({
worstErrorTool: tool({ tool: 'cohort-create', error_rate_pct: 6.3 }),
topTool: tool(),
})
expect(prompt).toContain('cohort-create')
expect(prompt).toContain('6.3%')
expect(prompt).toContain('$mcp_tool_call')
})
it('asks a trend question on the busiest tool when nothing errors', () => {
const prompt = buildMaxPrompt({
worstErrorTool: tool({ error_rate_pct: 0 }),
topTool: tool({ tool: 'read-data-schema' }),
})
expect(prompt).toContain('read-data-schema')
expect(prompt).toContain('trended')
})
})
describe('buildEditorPrompt', () => {
it('targets the detected client', () => {
expect(buildEditorPrompt({ client: 'Cursor' }).label).toBe('Paste this into Cursor')
expect(buildEditorPrompt({ client: 'OpenAI Codex' }).label).toBe('Paste this into Codex')
})
it('falls back to a generic target for unknown or missing clients', () => {
expect(buildEditorPrompt({ client: null }).label).toBe('Paste this into your agent')
expect(buildEditorPrompt({ client: 'Some New Agent' }).label).toBe('Paste this into your agent')
})
})
describe('buildChips', () => {
it('drops chips whose data is missing', () => {
expect(
buildChips({
topTool: null,
worstErrorTool: null,
kpis: { ...KPIS, p95LatencyMs: metric(0) },
client: null,
})
).toEqual([])
})
it('drops the flakiest chip when it is the same tool as the busiest', () => {
const chips = buildChips({
topTool: tool({ tool: 'execute-sql' }),
worstErrorTool: tool({ tool: 'execute-sql', error_rate_pct: 12 }),
kpis: KPIS,
client: null,
})
expect(chips.map((c) => c.key)).toEqual(['top-tool', 'p95'])
})
it('includes all four when data is present, in order', () => {
const chips = buildChips({
topTool: tool(),
worstErrorTool: tool({ tool: 'cohort-create', error_rate_pct: 12 }),
kpis: KPIS,
client: 'Cursor',
})
expect(chips.map((c) => c.key)).toEqual(['top-tool', 'worst-error', 'p95', 'client'])
expect(chips.find((c) => c.key === 'worst-error')?.tone).toBe('danger')
expect(chips.find((c) => c.key === 'client')?.harness).toBe('Cursor')
})
})
})