Deliver a structured speaking task from data/speaking.json, with hint scaffolding, target patterns, and a timer target — revealing the sample answer only after the learner attempts.
Invoke this skill when the learner runs /jp-speak or when generate-daily-tasks assigns a speaking task. During speaking+review weeks (17–24), this is the primary skill.
| Parameter | Type | Required | Description |
|---|---|---|---|
level | string | yes | Learner's JLPT level: "N5", "N4", "N3" |
topic | string | no | Topic preference. If omitted, auto-selected by recent focus. |
duration_target | integer | no | Target speaking duration in seconds. Used to filter items. |
item_id | string | no | Specific speaking item ID. If omitted, auto-selected. |
show_sample |
| boolean |
| no |
If true, reveal sample_answer_jp immediately. Default: false. |
A speaking session object:
| Field | Type | Description |
|---|---|---|
item_id | string | The speaking item used |
prompt_zh | string | Task description in Chinese |
duration_target_seconds | integer | Target duration |
hints | string[] | Scaffolding sentence starters (always shown) |
target_patterns | string[] | Grammar patterns the learner should try to use |
timer_guidance | string | Human-readable timer instruction |
sample_answer_jp | string | Model answer (withheld until learner submits, unless show_sample: true) |
follow_up_prompt | string | Optional follow-up question to extend the conversation |
data/speaking.json by level and topic (if provided).duration_target_seconds <= duration_target (if provided).storage/daily_state.json).Always show hints[] before the learner starts. Hints are sentence starters, not complete answers. Present them as a numbered list labeled "可以用這些句型開始：".
The sample_answer_jp is withheld by default. After the learner indicates they have finished (e.g., types their answer or says "done"), reveal the sample and recommend running /jp-review for personalized correction.
After the learner completes the task, generate one natural follow-up question in Japanese to extend the conversation (e.g., if they described their weekend plans, ask: 「それは楽しそうですね。どんな食べ物が好きですか？」).
Convert duration_target_seconds to a human-friendly instruction:
Output (abbreviated):
{
"item_id": "speak_001",
"prompt_zh": "用日語做一段30秒的自我介紹。包含：你的名字、你來自哪裡、你的職業或身份、你學日語的理由。",
"duration_target_seconds": 30,
"hints": [
"はじめまして、〜と申します。",
"〜から来ました。",
"〜をしています。",
"日本語を勉強している理由は〜です。"
],
"target_patterns": ["〜と申します", "〜から来ました", "〜をしています"],
"timer_guidance": "Try to speak for about 30 seconds — roughly 3–5 sentences.",
"follow_up_prompt": "どうして日本語を勉強し始めましたか？",
"sample_answer_jp": "[withheld until submitted]"
}