mooncake pilot run¶
Execute pilot iteration
Flags¶
| Flag | Type | Default | Description |
|---|---|---|---|
--goal / -g |
string | - | Goal description |
--plan / -p |
string | - | Path to plan YAML file |
--stdin |
bool | false | Read plan from stdin |
--provider |
string | - | LLM provider: anthropic-cli, anthropic-http, or openai-shape (omit for auto-discovery) |
--endpoint |
string | - | OpenAI-compatible /v1 base URL (e.g. http://localhost:11434/v1); required for --provider openai-shape unless MOONCAKE_PILOT_ENDPOINT is set |
--model |
string | - | Model name (when using --provider) |
--max-iterations |
int | 5 |
Maximum iterations for loop mode |
--auto-apply |
bool | false | Skip the plan-confirm gate (required for unattended/CI runs; spec-67 ยง10) |
--style |
string | - | Planning style: plan (single complete plan, default) or step (one action per turn). Overrides MOONCAKE_PILOT_STYLE. |