Phase 4: Copilot fake fixture and offline test suite

[tinkermonkey]

Create the fake-copilot.mjs fixture (if not already present from the feat/v0.4-copilot-backend merge) and a corresponding offline test file CliProcess.copilot.test.ts that exercises CliProcess with backend: 'copilot' against the fixture. This phase is blocked until the Copilot backend implementation is merged to main.

Requirements

• Prerequisite: feat/v0.4-copilot-backend must be merged to main before this phase begins
• If fake-copilot.mjs and CliProcess.copilot.test.ts are not brought in by the merge, create them:
  • fake-copilot.mjs driven by FAKE_SCENARIO env var, consuming ACP JSON-RPC requests from stdin, writing ACP-compliant NDJSON to stdout
  • Scenarios: golden-path (full handshake + one assistant turn + idle), stall, ignore-sigterm, nonzero-exit, permission-request
• CliProcess.copilot.test.ts must mirror the structure of the existing CliProcess.test.ts: inject fake-copilot.mjs into PATH as copilot binary via beforeAll, instantiate CliProcess with backend: 'copilot'
• Cover the same scenario categories as the Claude fake suite: golden path, stall (idle_timeout), SIGTERM/SIGKILL escalation, nonzero exit, AbortSignal (pre-flight and mid-run)
• ReadyEvent, TextEvent, and DoneEvent emitted by the Copilot backend must satisfy the same normalized field contracts as the Claude backend
• All tests must pass under npm test with no live binary or credentials

Design Guidance

Implements the 'Copilot Fake Fixture' (FR-4) and 'Copilot Fake Test Suite' (FR-5) from the business requirements, and the 'Extended Fake Fixture Scenarios' and 'PATH Injection for Fake Binaries' patterns from the architecture design. The ACP protocol handshake sequence for the golden-path scenario is specified in the architecture's component design.

Acceptance Criteria

• feat/v0.4-copilot-backend is merged to main before work begins
• fake-copilot.mjs exists under src/__tests__/fixtures/ with all five scenarios implemented
• FAKE_SCENARIO=golden-path reads ACP initialize from stdin and writes a valid ACP response sequence to stdout, then exits 0
• CliProcess.copilot.test.ts exists and covers golden path, stall, SIGTERM/SIGKILL, nonzero exit, and AbortSignal
• Collected events from the golden-path scenario include ReadyEvent, TextEvent, and DoneEvent with the same field contracts as the Claude backend
• All Copilot fake tests pass under npm test without the real copilot binary
• Code is reviewed and approved

Dependencies

Prerequisite: feat/v0.4-copilot-backend merged to main

[tinkermonkey]

Implementation

---

TextEvent only appears in a test description string, not as a type, so removing it from imports is correct. All tests pass cleanly.

---

Summary

What was done

fake-copilot.mjs (src/__tests__/fixtures/fake-copilot.mjs) — already existed with all five required scenarios: golden-path, stall, ignore-sigterm, nonzero-exit, and permission-request.

CliProcess.copilot.test.ts (src/__tests__/CliProcess.copilot.test.ts) — created as a standalone file mirroring CliProcess.test.ts's structure:

• beforeAll: injects fake-copilot.mjs into PATH as the copilot binary
• afterAll/afterEach: cleans up temp dir and FAKE_SCENARIO
• Covers all required scenario categories:
  • Golden path — ReadyEvent with ACP session ID, TextEvent with content, DoneEvent, ProgressEvent, strictly-increasing seq
  • Stall / idle_timeout — watchdog fires after 1s of silence
  • max_timeout — wall-clock ceiling
  • SIGTERM → SIGKILL escalation — ignore-sigterm scenario
  • Nonzero exit — nonzero-exit scenario
  • AbortSignal pre-flight — single ErrorEvent { code: 'aborted' }, no subprocess spawned (new test added)
  • AbortSignal mid-run — abort after ReadyEvent
  • permission/request — surfaces as RawEvent

CliProcess.test.ts — removed the now-duplicate describe('copilot backend', ...) block; unused TextEvent and FAKE_COPILOT_SRC imports also cleaned up.

Result: 5 test files, 92 tests, all passing under npm test with no live binary or credentials.

---

Generated by Orchestrator Bot 🤖
Processed by the senior_software_engineer agent

[tinkermonkey]

🔄 Starting Review Cycle

Reviewer: Code Reviewer
Maker: Senior Software Engineer
Max Iterations: 5

The automated maker-checker review cycle is now starting. The reviewer will evaluate the work, and if changes are needed, the maker will be automatically re-invoked with feedback.

---

Review cycle initiated by Switchyard

[tinkermonkey]

Code Review

---

No copilot references remain in the Claude test file. The cleanup is thorough.

Here is my review:

---

Status

APPROVED

Issues Found

Critical (Must Fix)

None

High Priority (Should Fix)

None

Low Priority (Observations — not blocking)

• Unused ErrorEvent type import: CliProcess.copilot.test.ts:7 imports ErrorEvent but never uses it as a type annotation or type assertion. It only appears in test description strings (which are plain string literals, not type references). The Claude test file uses ErrorEvent for explicit as ErrorEvent | undefined casts (lines 130, 184), but the copilot test skips those casts entirely. Since this is a type import it's erased at compile time and has zero runtime cost, but it's still dead code. Consider removing it or adding explicit type assertions to match the Claude test style.

• collect() signature diverges from Claude test: The copilot collect(extra: Partial<ProcessOptions> = {}) merges BASE internally, while the Claude test's collect(opts: ProcessOptions) takes the full options object. This is a deliberate convenience improvement that reduces boilerplate in each test call (await collect() vs await collect(BASE)). Noting it as an intentional style difference — both approaches are reasonable.

Positive Observations

• Clean extraction: The copilot describe block was fully removed from CliProcess.test.ts with no orphaned references. The unused TextEvent and FAKE_COPILOT_SRC imports were cleaned up. Grep confirms zero copilot references remain.

• Structural mirroring: The new file accurately mirrors the Claude test's structure: beforeAll PATH injection, afterAll cleanup, afterEach env reset, and scenario grouping (golden path → availability → error paths → timeouts → AbortSignal → extended scenarios).

• Coverage completeness: All six required scenario categories are covered: golden path (6 assertions including seq monotonicity, ReadyEvent/DoneEvent session IDs, TextEvent content), stall/idle_timeout, max_timeout, SIGTERM→SIGKILL escalation, nonzero exit, AbortSignal (both pre-flight and mid-run), plus the permission-request extended scenario.

• Test isolation: pool: 'forks' in vitest config ensures process.env.PATH mutations are isolated per test file. The fake binary names (claude vs copilot) don't collide even if isolation were weaker.

• Same field contracts verified: The golden-path tests assert ReadyEvent.sessionId and DoneEvent.sessionId carry the ACP session ID. The model and tools fields are optional in the ReadyEvent interface, and usage is optional in DoneEvent, so their absence in copilot events correctly reflects the ACP protocol's simpler data model while satisfying the same TypeScript contracts.

Source Artefact Updates

None

---

Generated by Orchestrator Bot 🤖
Processed by the code_reviewer agent

[tinkermonkey]

Review Cycle Complete

Status: APPROVED

Iterations: 1/5

Maker: Senior Software Engineer

Reviewer: Code Reviewer

Review approved after 1 iteration(s)

---

Automated review cycle by Switchyard

[tinkermonkey]

Starting Repair Cycle (Testing)

Agent: Senior Software Engineer

Test Types: compilation, unit, integration, ci

Max Iterations: 100

Container: Isolated Docker container
Branch: feature/issue-2-add-full-suite-of-live-integra

The automated test-fix-validate cycle is now starting in a containerized environment. Tests will be run, and if failures are detected, the agent will automatically fix them and re-run tests until all tests pass.

---

Repair cycle initiated by Switchyard

[tinkermonkey]

✅ Repair Cycle Complete

Container: repair-cycle-code-wrapper-7-4d984e8a

Exit Code: 0

Total Agent Calls: 4

Duration: 109.7s

• compilation: ✅ PASSED (1 iterations)
• unit: ✅ PASSED (1 iterations)
• integration: ✅ PASSED (1 iterations)
• ci: ✅ PASSED (1 iterations)

---

Repair cycle executed by Switchyard (containerized)