Orca is the place for working with a fleet of agents (e.g. Grok Bot, Claude Code). Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.
TypeScript adeagent-ideai-agentsautomationbotclaude-codecodexghosttygrok-bothermes-agentmobile-appmuseopenclawopencodeorchestrationpiruntimeterminalworktreesyc-backed
import { describe, expect, it } from 'vitest'
import { agentSessionFailureSentence } from './agent-session-failure-words'
describe('agent-specific sign-in guidance', () => {
it.each([
['Missing key', 'Missing key. Then send your message again.'],
['Missing key.', 'Missing key. Then send your message again.'],
['Missing key!', 'Missing key! Then send your message again.'],
['Missing key?', 'Missing key? Then send your message again.'],
['"Missing key."', '"Missing key." Then send your message again.'],
['(Missing key.)', '(Missing key.) Then send your message again.'],
["('Missing key!')", "('Missing key!') Then send your message again."],
['“Missing key?”', '“Missing key?” Then send your message again.'],
['Missing key。', 'Missing key。Then send your message again.'],
['Missing key!', 'Missing key!Then send your message again.'],
['Missing key?', 'Missing key?Then send your message again.']
])('separates guidance after the provider detail %j', (detail, tail) => {
expect(
agentSessionFailureSentence(
{ kind: 'notSignedIn', detail: { text: detail, audience: 'person' } },
'rejection',
{ agentName: 'Pi' }
)
).toBe(
`Sign in to Pi by running \`pi\` and using \`/login\` on the computer running this chat. ${tail}`
)
})
it('leaves an unpunctuated detail alone when no guidance follows it', () => {
const fact = {
kind: 'notSignedIn',
detail: { text: 'No API key found for anthropic', audience: 'person' }
} as const
const row = agentSessionFailureSentence(fact, 'row', { agentName: 'Pi' })
expect(row.endsWith(fact.detail.text)).toBe(true)
expect(
agentSessionFailureSentence(fact, 'rejection', { agentName: 'Pi', retryControl: true })
).toBe(row)
})
it.each([
['Claude', 'claude auth login'],
['Codex', 'codex login'],
['Grok', 'grok login'],
['OpenCode', 'opencode auth login'],
['Pi', 'pi'],
['OMP', 'Sign in to OMP.']
])(
'names %s and its verified command without inventing account selection',
(agentName, command) => {
const fact = {
kind: 'notSignedIn',
detail: { text: 'The configured provider has no API key.', audience: 'person' }
} as const
const row = agentSessionFailureSentence(fact, 'row', { agentName })
const rejection = agentSessionFailureSentence(fact, 'rejection', { agentName })
expect(row).toContain(command)
expect(rejection).toContain(command)
expect(row).toContain(fact.detail.text)
expect(rejection).toContain(fact.detail.text)
expect(row).not.toContain('send your message again')
expect(row).not.toContain('selected account')
if (agentName === 'Pi') {
expect(row).toContain('`/login`')
}
}
)
it('keeps diagnostic text literal and keeps log-only detail off the visible sentence', () => {
const detail = 'No API key found for {{agent}} $t(fake.key)'
expect(
agentSessionFailureSentence(
{ kind: 'notSignedIn', detail: { text: detail, audience: 'person' } },
'rejection',
{ agentName: 'Pi' }
)
).toContain(detail)
expect(
agentSessionFailureSentence(
{ kind: 'notSignedIn', detail: { text: detail, audience: 'log' } },
'row',
{ agentName: 'Pi' }
)
).not.toContain(detail)
})
it('keeps command retry and managed-account advice', () => {
expect(
agentSessionFailureSentence({ kind: 'notSignedIn', account: 'managed' }, 'row', {
agentName: 'Claude'
})
).toBe("This Claude account isn't signed in. Sign in again in Claude Accounts settings.")
expect(
agentSessionFailureSentence({ kind: 'notSignedIn', account: 'managed' }, 'rejection', {
agentName: 'Codex',
command: 'clear'
})
).toBe(
"This Codex account isn't signed in. Sign in again in Codex Accounts settings. Run /clear again."
)
})
})