AIResearch Plugin

Six-lens equity research inside your coding agent. One command, one integrated report — no lens-picking, no chat limits, powered by whatever model you already pay for.

Lenses: Supply Chain & Value Flow · Fundamentals · Macro · Technicals · Sentiment & Positioning · Risk Matrix & Conviction

Install — Claude Code

/plugin marketplace add WOOK98/airesearch-plugin
/plugin install airesearch@airesearch-marketplace

Then reload plugins (or restart Claude Code) and run the namespaced commands:

/reload-plugins
/airesearch:deep-dive NVDA          # full six-lens report
/airesearch:deep-dive AXTI risk     # single-lens focus
/airesearch:deep-dive "Humanoid robot"  # industry mode: value chain + constituent universe
/airesearch:snapshot MU             # 30-second read
/airesearch:morning-brief           # watchlist triage

The skills also trigger implicitly — just ask "give me a deep dive on LITE" or "quick take on SNDK?".

No API key needed to start — skills work with web search out of the box. For live market data, see Pro data layer.

Install — ChatGPT / Codex Personal Marketplace

The repository ships a .codex-plugin/plugin.json manifest and marketplace assets. See docs/codex-install.md for the personal marketplace JSON and local install steps.

After install, the same skill prompts are available from the plugin card.

Install — DeepSeek Harness

AIResearch also ships a DSH bundle. It registers four skills in the Harness skill catalog: airesearch-snapshot, airesearch-deep-dive, airesearch-morning-brief, and airesearch-filing.

npx @dsh-1024store/cli add WOOK98/airesearch-plugin --profile web

After installation, restart the web profile if it is already running. Then ask naturally:

snapshot TSLA
deep dive Humanoid robot
morning brief NVDA, TSLA, 0700.HK
filing NVDA latest 10-K

DeepSeek harness adapter

For non-Claude/Codex runners, this repository also ships a prompt-only DeepSeek harness bundle under deepseek-harness/. It contains a task manifest, shared system prompt, task prompts, and a zero-dependency Python runner: