完整 Git 提交记录

refactor: split calculation and llm workers

fix: reduce analysis worker log noise

refactor: move python workers into scripts

docs: plan mentor onboarding and report grants

fix: align current architecture conventions

refactor: formalize vision task api routes

Update vision diagnosis prompt protocol

Align vision report status response

test

Fix vision worker MySQL authentication

Simplify vision diagnosis prompt output

Optimize MySQL indexes

Simplify mini program auth routes

Simplify mini program access tokens

Add mini program token authentication

Add Douyin mini program silent login

Make payment prices env configurable

Support WeChat Pay platform public key IDs

Improve WeChat Pay notification diagnostics

Resolve WeChat Pay certificate paths

Add ICP filing to footer

Improve OCR demo upload handling

Restrict OCR vision LLM to responses API

Update OCR vision prompt and responses payload

Store OCR demo tasks in database

Use user scoped OSS keys for mini program screenshots

Store mini program screenshots in OSS

Use direct vision LLM for screenshot demo

Validate OCR screenshot input

Expose structured OCR polling data

Use text LLM for OCR report generation

Split OCR extraction from creator report generation

Store mini program OCR API results

Only log OCR demo worker queued tasks

Silence OCR demo worker idle polling

Adapt OCR API for Aliyun Qwen OCR

Add OCR API pre-processing step

Add structured OCR extraction results

Add staged OCR progress updates

Register OCR prompt template

Log OCR vision LLM requests

Record mini program OCR upload tasks

Compress OCR images before vision requests

Align OCR image size limit

Support base64 OCR API uploads

Add OCR demo API routes

Improve OCR demo worker logging

Require explicit OCR vision provider

text

Fix report history status for LLM failures