Lobster Research: automation for Hermes Agent

Skill that generates finance and industry research reports as PDFs, with data collection scripts

What Lobster Research does

Lobster Research (龙虾调研) pairs deterministic data collection in Python with analysis written by the host agent. Users ask in natural language and a two-layer keyword router picks the domain and output type. There are three outputs: a text flash update, a 3 to 5 page PDF brief, and an 8 to 15 page PDF research report. It covers 23 domains, including A-shares, Hong Kong and US stocks, ETFs, commodities and futures, plus 12 non-financial areas such as technology, gaming and biomedicine.

Work runs in phases. Phase 1 collects quotes, candlestick data with technical indicators, index data and web search results from sources including Sina Finance, Tencent Finance, AKShare and several search engines. Phase 2 has the agent fill a JSON report input that follows a template structure. Phase 3 renders the PDF with selectable palettes, render types and layouts. The repository ships 29 report templates and a --type quick mode that skips slow collection steps. A simulated-portfolio phase records trade decisions and reflections after stock analysis.

Key features

Natural-language routing to one of 23 domains and three output types
Data from Sina Finance, Tencent Finance, AKShare and web search engines
29 report templates: 7 briefs and 22 research reports
PDF styling with a 10-color palette and solid, gradient and liquid render types
A --type quick mode that skips slow data collection steps
Simulated portfolio tracking with trade decisions and reflection

When to use it

Scheduling a market check or a holdings snapshot through an agent
Producing a structured industry report with citations and data tables
Tracking technology or policy trends with periodic briefs

Who it is for: Investors, analysts and researchers who want structured research reports produced by an AI agent.

How it fits with Hermes Agent

The repository is tagged hermes-skill, and the README describes the skill as OpenClaw-compatible and usable in any agent that can run Python scripts and read and write JSON.

How to install Lobster Research

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

git clone https://github.com/hegeo/lobster-research.git
cd lobster-research
pip install -r requirements.txt

Requirements: Python, an agent platform that can run Python scripts and read and write JSON, and optionally search API keys such as Tavily or Bing in config/settings.json.

Note: The README is bilingual in Chinese and English, describes the skill as OpenClaw-compatible, and GitHub reports the license as NOASSERTION.

FAQ

What is Lobster Research?

Lobster Research is an agent skill that generates flash updates, short briefs and in-depth research reports. It collects market and web data with code, then the agent writes the analysis and a PDF is rendered.

Does Lobster Research work with Hermes Agent?

The repository carries a hermes-skill topic, and the README says any agent that can run Python scripts, read and write JSON and deliver attachments is compatible. It names WorkBuddy, QClaw and OpenClaw as the main platforms.

How do I install Lobster Research?

Send the repository link to your agent and ask it to install the skill, or clone the repository and run pip install -r requirements.txt. Search API keys can be added by copying config/settings.json.example to config/settings.json.