Files
orca/src/shared/codex-subagent-roster.test.ts
T
Brennan Benson b4d732685c feat(agent-status): combine Codex child work through the shared main-agent status fold (#22475)
* feat(agent-status): combine Codex child work through the shared main-agent status fold

* docs(agent-status): correct two comments the waiting child-work arm made stale

A child failure reported in place as `blocked` now pins the row `waiting`, not
`working`; and no relay ever sent an unfolded `working` beside a waiting child.

* fix(agent-status): only a waiting child asks for a human

A child's `blocked` state means its task failed (the only producer maps a
failed background task to it, and the background-task view labels it
"failed"), not that a human must act. Folding it into the waiting arm would
surface a failed child as needs-you. It stays live work, as before this
series.

* docs(agent-status): say a waiting child, not a blocked one, makes the row wait

A child's blocked state means it failed; only its waiting state feeds the
waiting arm. Two fold comments, a test describe and two parity story names
still called the waiting child blocked.

* docs(agent-status): name where a child's wait is still lost, and pin the structured lane's real input

The doc said the Claude hook lane's rows match Codex and that every lane feeds a
child's wait into the fold. Neither holds: Claude keeps the wait in one slot the
next main agent event overwrites, the structured lane turns a child's prompt
into the main agent's own attention, and Codex drops its roster on a root Stop
when it tracks no child transcripts. The parity story now drives the structured
lane with the input it actually receives.
2026-09-23 22:09:07 -07:00

131 lines
4.6 KiB
TypeScript

import { describe, expect, it } from 'vitest'
import {
AGENT_MODEL_MAX_LENGTH,
AGENT_STATUS_MAX_SUBAGENTS,
AGENT_TYPE_MAX_LENGTH
} from './agent-status-types'
import {
codexRosterChildWorkLiveness,
codexRosterToSnapshots,
finishCodexSubagent,
setCodexSubagentModel,
upsertCodexSubagent,
type CodexSubagentRoster
} from './codex-subagent-roster'
describe('Codex subagent roster', () => {
it('normalizes retained identity fields before storing them', () => {
const roster: CodexSubagentRoster = new Map()
upsertCodexSubagent(
roster,
' child-1 ',
{
agentType: `reviewer\n${'x'.repeat(AGENT_TYPE_MAX_LENGTH * 2)}`,
description: 'Review the sidebar lifecycle',
model: `gpt-model-${'x'.repeat(AGENT_MODEL_MAX_LENGTH * 2)}`,
state: 'working'
},
10
)
const snapshot = codexRosterToSnapshots(roster)?.[0]
expect([...roster.keys()]).toEqual(['child-1'])
expect(snapshot?.agentType).toHaveLength(AGENT_TYPE_MAX_LENGTH)
expect(snapshot?.agentType).not.toContain('\n')
expect(snapshot?.description).toBe('Review the sidebar lifecycle')
expect(snapshot?.model).toHaveLength(AGENT_MODEL_MAX_LENGTH)
finishCodexSubagent(roster, ' child-1 ')
expect(roster.size).toBe(0)
})
it('rejects an id that would normalize to an invisible child', () => {
const roster: CodexSubagentRoster = new Map()
upsertCodexSubagent(roster, ' ', { state: 'waiting' }, 10)
expect(roster.size).toBe(0)
})
it('bounds live storage while admitting a replacement after one child stops', () => {
const roster: CodexSubagentRoster = new Map()
for (let index = 0; index <= AGENT_STATUS_MAX_SUBAGENTS; index += 1) {
upsertCodexSubagent(roster, `child-${index}`, { state: 'working' }, index)
}
expect(roster.size).toBe(AGENT_STATUS_MAX_SUBAGENTS)
expect(roster.has(`child-${AGENT_STATUS_MAX_SUBAGENTS}`)).toBe(false)
finishCodexSubagent(roster, 'child-0')
upsertCodexSubagent(roster, 'replacement', { state: 'working' }, 100)
expect(roster.size).toBe(AGENT_STATUS_MAX_SUBAGENTS)
expect(roster.has('replacement')).toBe(true)
})
describe('codexRosterChildWorkLiveness', () => {
it('reads every child as agent work, a waiting one above the rest, and nothing as a watch loop', () => {
const roster: CodexSubagentRoster = new Map()
expect(codexRosterChildWorkLiveness(undefined)).toBeNull()
expect(codexRosterChildWorkLiveness(roster)).toBeNull()
upsertCodexSubagent(roster, 'a', { state: 'working' }, 1)
expect(codexRosterChildWorkLiveness(roster)).toBe('working')
upsertCodexSubagent(roster, 'b', { state: 'waiting' }, 2)
expect(codexRosterChildWorkLiveness(roster)).toBe('waiting')
finishCodexSubagent(roster, 'b')
expect(codexRosterChildWorkLiveness(roster)).toBe('working')
})
})
describe('setCodexSubagentModel', () => {
it('records the model without disturbing the child lifecycle or label', () => {
const roster: CodexSubagentRoster = new Map()
upsertCodexSubagent(roster, 'child-1', { description: '/root/audit', state: 'waiting' }, 10)
setCodexSubagentModel(roster, 'child-1', ' gpt-5.6-terra ')
expect(codexRosterToSnapshots(roster)).toEqual([
{
id: 'child-1',
agentType: undefined,
description: '/root/audit',
model: 'gpt-5.6-terra',
state: 'waiting',
startedAt: 10
}
])
})
it('never creates a row for a child that is no longer tracked', () => {
const roster: CodexSubagentRoster = new Map()
upsertCodexSubagent(roster, 'child-1', { state: 'working' }, 10)
finishCodexSubagent(roster, 'child-1')
// A model read racing a completed child must not resurrect its row.
setCodexSubagentModel(roster, 'child-1', 'gpt-5.6-terra')
expect(roster.size).toBe(0)
})
it('keeps a known model when the new value is empty', () => {
const roster: CodexSubagentRoster = new Map()
upsertCodexSubagent(roster, 'child-1', { model: 'gpt-5.6-sol', state: 'working' }, 10)
setCodexSubagentModel(roster, 'child-1', ' ')
setCodexSubagentModel(roster, 'child-1', undefined)
expect(roster.get('child-1')?.model).toBe('gpt-5.6-sol')
})
it('bounds an oversized model to the shared cap', () => {
const roster: CodexSubagentRoster = new Map()
upsertCodexSubagent(roster, 'child-1', { state: 'working' }, 10)
setCodexSubagentModel(roster, 'child-1', 'x'.repeat(AGENT_MODEL_MAX_LENGTH + 50))
expect(roster.get('child-1')?.model).toHaveLength(AGENT_MODEL_MAX_LENGTH)
})
})
})