PY7 P7-A12-WP1 — Define TestCodexBuildFoodTruckCmd

[Trecek]

Summary

Add a TestCodexBuildFoodTruckCmd test class to tests/execution/backends/test_codex_backend.py that validates the command structure, environment variables, flag presence/absence, and resume-path behavior of CodexBackend.build_food_truck_cmd. The class uses a BASE class-attribute dict for required kwargs, an autouse _clean_env fixture to strip campaign/kitchen env vars, and ~20 focused test methods covering structural assertions, env var injection, and resume subcommand placement.

Requirements

Goal

Define TestCodexBuildFoodTruckCmd with BASE fixture, _clean_env, and structural/flag tests for build_food_truck_cmd.

Context

• Phase: Codex Food Truck and Per-Step Backend Mixing (Milestone: 7-codex-food-truck-and-per-step-backend-mixing)
• Assignment: P7-A12 (Add unit tests for CodexBackend.build_food_truck_cmd command structure (sandbox read-only, web_search disabled, env vars, resume subcommand path))
• Depends on: P7-A3-WP1
• Depended on by: P8-A6-WP1

Deliverables

• TestCodexBuildFoodTruckCmd class with BASE fixture, _clean_env, and structural tests
• tests/execution/backends/test_codex_backend.py — env var test methods in TestCodexBuildFoodTruckCmd
• tests/execution/backends/test_codex_backend.py — resume-path test methods in TestCodexBuildFoodTruckCmd

Acceptance Criteria

• spec.cmd[0] == 'codex'
• spec.cmd[1] == 'exec'
• '--json' in spec.cmd
• '--sandbox' followed by 'read-only'
• '-c' followed by 'web_search=disabled'
• '-a' followed by 'never'
• '--add-dir' not in spec.cmd
• '--plugin-dir' not in spec.cmd
• prompt is last token
• isinstance(spec, CmdSpec)
• isinstance(spec.cmd, tuple)
• AUTOSKILLIT_HEADLESS == '1'
• AUTOSKILLIT_SESSION_TYPE == SESSION_TYPE_ORCHESTRATOR
• CAMPAIGN_ID present/absent based on env
• KITCHEN_SESSION_ID present/absent based on env
• AUTOSKILLIT_COMPLETION_MARKER set correctly
• AUTOSKILLIT_PROVIDER_PROFILE absent when not supplied
• 'resume' present when resume_session_id supplied
• session_id follows 'resume'
• '--json' always present
• 'resume' absent when resume_session_id is None

Closes #2906

Implementation Plan

Plan file: /home/talon/projects/autoskillit-runs/impl-20260525-222057-105440/.autoskillit/temp/make-plan/py7_p7_a12_wp1_plan_2026-05-25_222500.md

🤖 Generated with Claude Code via AutoSkillit

Token Usage Summary

Step	Model	count	uncached	output	cache_read	peak_ctx	turns	cache_write	time
plan	claude-sonnet-4-6	1	62	13.0k	1.7M	97.9k	139	73.4k	12m 31s
verify	claude-sonnet-4-6	1	148	11.1k	815.8k	59.9k	49	45.6k	3m 37s
implement*	MiniMax-M2.7-highspeed	1	562.5k	6.3k	589.8k	25.8k	53	16.1k	2m 36s
prepare_pr*	MiniMax-M2.7-highspeed	1	72.2k	3.9k	208.6k	31.1k	23	23.9k	1m 25s
compose_pr*	MiniMax-M2.7-highspeed	1	58.9k	1.9k	203.3k	25.8k	16	15.2k	53s
review_pr	claude-sonnet-4-6	1	164	37.7k	961.8k	81.6k	58	68.1k	7m 54s
resolve_review	claude-sonnet-4-6	1	267	16.5k	1.7M	69.2k	74	55.2k	6m 19s
Total			694.2k	90.4k	6.2M	97.9k		297.4k	35m 16s

* Step used a non-Anthropic provider; caching behavior may differ.

Token Efficiency

Step	LoC Changed	cache_read/LoC	cache_write/LoC	output/LoC
plan	0	—	—	—
verify	0	—	—	—
implement	150	3931.9	107.2	41.8
prepare_pr	0	—	—	—
compose_pr	0	—	—	—
review_pr	0	—	—	—
resolve_review	0	—	—	—
Total	150	41217.1	1982.7	602.9

Model Usage Breakdown

Model	steps	uncached	output	cache_read	cache_write	time
claude-sonnet-4-6	4	641	78.3k	5.2M	242.3k	30m 22s
MiniMax-M2.7-highspeed	3	693.5k	12.2k	1.0M	55.1k	4m 53s