autoship-setup
Interactive setup wizard for AutoShip on OpenCode — model selection, tool detection, concurrency tuning
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Interactive setup wizard for AutoShip on OpenCode — model selection, tool detection, concurrency tuning
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Delegate GitHub issue-to-PR orchestration to the AutoShip OpenCode plugin.
Implementer role — writes code to implement features or fixes
AutoShip orchestration for OpenCode-only workers
Display current AutoShip orchestration status — running agents, quota bars, progress, token usage
Post-completion verification pipeline for OpenCode — review, simplify, re-verify, PR creation, CI monitoring, cleanup
Lead role — coordinates multiple agents and manages orchestration flow
| name | autoship-setup |
| description | Interactive setup wizard for AutoShip on OpenCode — model selection, tool detection, concurrency tuning |
| platform | opencode |
| tools | ["Bash","question","Write"] |
Guide users through model selection and configuration on first run.
/autoship-setup
/autoship-setup --no-tui \
--max-agents 10 \
--labels "agent:ready,autoship:in-progress" \
--refresh-models \
--orchestrator-model provider/model-a \
--reviewer-model provider/model-b
gh auth statusopencode modelscommand -v opencode >/dev/null 2>&1 && opencode --version
Ask user: "OpenCode not found. Install it first."
gh auth status
Ask user if not authenticated:
GitHub authentication required.
Run: gh auth login
Or set GH_TOKEN environment variable.
opencode models
Ask the user:
Which model configuration?
◉ Choose orchestrator and reviewer role models
└ They can be the same model or different models
└ Prefer capable free or OpenCode Go Kimi/Kimmy/Ling 2.6-family models when available
◉ Free-first OpenCode
└ Prefer configured free OpenCode models
└ Do not include paid models by default
◯ Custom OpenCode models
└ Operator chooses model IDs from the current `opencode models` output
└ Writes `.autoship/model-routing.json`
└ Explicitly selected non-free models are allowed
Ask:
How many concurrent agents?
◉ Conservative (5 agents)
└ Lowest cost, good for testing
◯ Standard (10 agents)
└ Balanced throughput/cost
◯ Aggressive (15 agents)
└ Higher parallelism for trusted queues
Ask:
Which labels should AutoShip monitor for issues?
◉ agent:ready (default)
└ Standard AutoShip queue label
◯ Custom labels (comma-separated)
└ e.g., "agent:ready,needs-triage"
Ask:
How often should AutoShip refresh model inventory?
◉ Auto-refresh on startup
└ Checks for new models each time AutoShip starts
◯ Manual only
└ Only refreshes when AUTOSHIP_REFRESH_MODELS=1 is set
# Interactive mode
AUTOSHIP_HOME="${OPENCODE_CONFIG_DIR:-${XDG_CONFIG_HOME:-$HOME/.config}/opencode}/.autoship"
bash "$AUTOSHIP_HOME/hooks/opencode/setup.sh"
# Non-interactive mode (--no-tui)
bash "$AUTOSHIP_HOME/hooks/opencode/setup.sh" --no-tui \
--max-agents 10 \
--labels "agent:ready" \
--refresh-models \
--orchestrator-model provider/model-a \
--reviewer-model provider/model-b
Setup preserves existing .autoship/model-routing.json by default so operators can edit it manually. Use AUTOSHIP_REFRESH_MODELS=1 or --refresh-models to regenerate free defaults from the current OpenCode model inventory.
✓ AutoShip configured!
Runtime: OpenCode
GitHub: Authenticated
Models: <free discovered models or explicit operator selection>
Concurrency: <N> agents
Labels: <configured labels>
Refresh: <auto/manual>
Next: Run /autoship to start
| Flag | Description | Default |
|---|---|---|
--max-agents N | Set max concurrent agents | 15 |
--labels LABEL1,LABEL2 | Comma-separated labels to monitor | agent:ready |
--refresh-models | Force refresh model inventory | false |
--planner-model MODEL | Set planner/coordinator/orchestrator/reviewer/lead model | best available role model |
--orchestrator-model MODEL | Set orchestrator model | prompted on first run |
--reviewer-model MODEL | Set reviewer model | prompted on first run |
--worker-models MODEL1,MODEL2 | Comma-separated worker models | auto-detect free models |
| Error | Recovery |
|---|---|
| OpenCode not found | Install OpenCode before starting workers |
| GitHub not authenticated | Run gh auth login or set GH_TOKEN |
| Config write fails | Check file permissions |
| No free models found | Use --worker-models to specify explicitly |
# Remove onboarding marker to re-run wizard
rm .autoship/.onboarded
# Or re-run with --no-tui flags
/autoship-setup --no-tui --max-agents 10 --refresh-models