ReMe/cookbook/auto-fin

Auto Fin Cookbook

中文

Auto Fin fetches a rolling window of CLS telegraph news (24 hours by default), selects items related to configured topics, searches ReMe for useful historical context, and writes one Chinese Markdown report with validated wikilinks. Current news and topic selection stay in runtime memory; only the final report becomes durable memory. The implementation lives in reme/steps/cookbook/auto_fin/ and is assembled by daily_cookbook.yaml.

Auto Fin has no reliable market-price feed. It does not calculate returns, targets, or entry points and is not investment advice.

Quick start

python -m pip install -e ".[core]"
export LLM_API_KEY="your-api-key"
export LLM_MODEL_NAME="qwen3.7-plus"
export LLM_BASE_URL="https://your-provider.example/v1"
reme start config=daily_cookbook job=auto_fin

LLM_MODEL_NAME defaults to qwen3.7-plus. There is no built-in LLM_BASE_URL, so set the OpenAI-compatible endpoint required by the selected provider.

The default topics are 黄金,机器人,半导体. Override them per run:

reme start config=daily_cookbook job=auto_fin topics="黄金,AI,存储芯片"

An empty value also uses the defaults.

Pipeline

CLS public telegraph endpoint (rolling 24 hours)
        ↓
normalize and deduplicate in RuntimeContext
        ↓
topic Agent selects real news IDs in bounded batches
        ↓
research Agent uses memory_search + read on historical memory
        ↓
validate historical wikilinks in code
        ↓
daily/YYYY-MM-DD/auto_fin.md

auto_fin_data_step signs and paginates the same endpoint used by the CLS website. It starts at the decision time and stops only after covering the exact preceding 24 hours. Requests are rate-limited and retried; malformed records and records outside the window are discarded.

auto_fin_topic_step receives batches of current news and returns only related news_id values. Code ignores unknown IDs and deduplicates repeated IDs, then preserves the source-news order. If nothing is relevant, the job succeeds as a skip without writing or sending a report.

auto_fin_merge_step receives only selected current news. It exposes memory_search and read, instructs the Agent to search no later than yesterday, and keeps current CLS IDs, times, and titles as plain evidence. The prompt limits wikilinks to historical Markdown actually used by the Agent; the code-level boundary independently keeps only existing, workspace-relative Markdown targets. Missing, absolute, escaping, backslash, and self-referential targets are degraded to their readable aliases.

Same-day reruns use the existing report as context and replace it with the revised result. The final write is atomic and refreshes the daily index. No JSONL, intermediate Markdown, or structured Agent output is written.

Parameters

Parameter	Default	Purpose
date	""	Empty uses today in Shanghai; an explicit value must equal today
now	""	Optional ISO 8601 decision time for testing or replay
topics	"黄金,机器人,半导体"	Comma-separated topics; empty also uses these defaults
window_hours	24	Rolling number of hours of CLS telegraph news to fetch; must be positive
request_interval	10	Minimum delay in seconds after every CLS request attempt; may be zero
max_retries	3	Maximum attempts for each CLS page request; must be at least one

The built-in schedules run daily at 09:30, 11:30, and 18:00 in Asia/Shanghai.

Output

reme_workspace/daily/YYYY-MM-DD/auto_fin.md

The report includes a title, description, current CLS evidence, historical analysis, contextual wikilinks, and a fixed non-investment disclaimer. Network errors and invalid Agent output fail explicitly; no relevant current news is a successful skip. If DINGTALK_CONVERSATION_IDS is empty, delivery is a no-op. If it is set, the DingTalk credentials described in the Daily Paper cookbook are required.

Validation

pytest tests/unit/test_auto_fin.py -v

Unit tests mock the CLS and Agent boundaries and do not contact external services.