fix: require write_file for WeChat immediate-context page requests
Memind CI / Test, build, and release guards (push) Failing after 12m9s
Memind CI / Test, build, and release guards (push) Failing after 12m9s
Skip fresh thumbnail forcing and retry in-session when load_skill completes without HTML落盘, so「把刚才的诗做成页面」can deliver the prior poem as a static page. Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
@@ -32,8 +32,11 @@ export function buildPageGenerateAgentPrompt(
|
||||
const immediateContextBlock = preferImmediateContext
|
||||
? [
|
||||
'【即时上下文优先】当前需求是对本会话紧邻内容的续作。',
|
||||
'“这个/刚才/做成页面”等指代只能解析为当前会话最近一轮内容,禁止用长期记忆、历史偏好或旧任务替换主题。',
|
||||
'“这个/刚才/做成页面”等指代只能解析为当前会话最近一条 assistant 回复中的完整正文(如诗全文、文章全文),禁止用长期记忆、历史偏好或旧任务替换主题。',
|
||||
'如果当前会话没有可辨识的紧邻内容,必须请用户补充主题,禁止猜测并生成其他页面。',
|
||||
'【write_file 硬门槛】必须先 `load_skill` → `static-page-publish`,再用 `write_file`/`edit_file` 把紧邻正文完整写入 `public/*.html` 后才能结束。',
|
||||
'禁止只调用 load_skill、todo_write 或 generate_image 就回复;没有 write_file 落盘 HTML 视为未完成。',
|
||||
'缩略图/配图可选:优先完成 HTML 落盘与 mindspace-cover;没有 hero 图时 cover 可用 CSS 渐变或 omit,不要卡在 generate_image。',
|
||||
'',
|
||||
].join('\n')
|
||||
: '';
|
||||
@@ -84,6 +87,19 @@ export function buildPageGenerateAgentPrompt(
|
||||
.join('\n');
|
||||
}
|
||||
|
||||
export function buildImmediateContextPageRepairPrompt(intent) {
|
||||
const topic = String(intent?.agentText ?? intent?.displayText ?? '').trim();
|
||||
return [
|
||||
'【即时上下文页面补写】上一轮页面生成未完成:已 load_skill 但未 write_file 落盘 HTML。',
|
||||
'请立即读取本会话最近一条 assistant 回复中的完整正文,将其排版为完整静态页。',
|
||||
'步骤:确认 static-page-publish 已加载 → 用 write_file 写入 public/*.html(含 mindspace-cover、platform-brand 页脚)→ 确认落盘后再回复唯一正式链接。',
|
||||
'禁止再次只 generate_image 或只回复文字;本轮必须完成 write_file。',
|
||||
topic ? `用户原话:${topic}` : '',
|
||||
]
|
||||
.filter(Boolean)
|
||||
.join('\n');
|
||||
}
|
||||
|
||||
export function buildPagePublishFailureText({
|
||||
missingSharePreview = false,
|
||||
missingPlatformBrand = false,
|
||||
|
||||
Reference in New Issue
Block a user