How to use the Market Sentiment Analysis agent skill

What the sentiment-analysis skill makes the agent do

An agent skill is a folder containing a SKILL.md file with instructions the agent reads when a request matches the skill's description. This one covers five sentiment dimensions: fear-and-greed indicators, put-call ratios, margin financing and securities lending, northbound capital, and social-media sentiment.

The skill treats extreme sentiment readings as contrarian signals and includes a template for combining the dimensions into a composite 0-100 score with suggested position sizes. It is mainly reference knowledge and output templates. It does not fetch live market data.

• Interprets the crypto Fear & Greed Index bands from 0 (extreme fear) to 100 (extreme greed).
• Explains A-share substitutes for a fear-and-greed index, such as turnover rate and limit-up/limit-down counts.
• Explains put-call ratio ranges and how to read them alongside VIX.
• Describes margin balance and northbound (Stock Connect) signals with example thresholds.
• Provides a social-media sentiment checklist and a report template with a sentiment dashboard, score, interpretation and risk notes.

Install the sentiment-analysis skill

Download the skill ZIP from the page you are reading. Then choose one of two methods depending on how you use your agent.

For Claude Code, unzip the file into your personal skills folder ~/.claude/skills/ or into a project-level .claude/skills/ folder inside your project. The skill folder should contain SKILL.md.

In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal install: ~/.claude/skills/ (available across your projects).
• Project install: .claude/skills/ inside a specific project.
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP.

Trigger the skill with natural requests

You do not need a special command. The agent loads the skill when your request is about market sentiment. Name the market and the indicator you care about to make the match more likely.

Ask for a full report when you want the dashboard template, or ask about one indicator when you only need an interpretation.

• Give me a sentiment report for A-shares: turnover, margin balance and northbound flow.
• What does a put-call ratio of 1.6 mean for the 50ETF? Is it contrarian bullish?
• Is crypto currently in extreme greed territory based on the fear-and-greed bands?
• Build a composite sentiment score from these numbers I paste in, with a position-size suggestion.
• Draft a social-media sentiment checklist for tracking a stock on Xueqiu and Reddit.

Read the output: examples of what you get

When you ask for a full analysis, the skill's template produces a sentiment dashboard table, a composite score with its band, an interpretation, an operating suggestion and a risk note. The example values in the template are illustrative, so replace them with figures you have checked yourself.

The composite score uses weights of 0.25 for fear-and-greed, 0.20 for put-call ratio, 0.20 for margin, 0.20 for northbound and 0.15 for social media. Each dimension is first normalised to 0-100. A score of 80-100 maps to a suggested 0-20% position.

What you need before you start

The skill lists no requirements and runs no scripts. It does not use the network. Its usefulness depends on the data you give the agent, because the file contains thresholds and examples rather than current figures.

Some data the framework mentions may require paid services, for example Bloomberg sentiment indicators or Glassnode. Check the source of each figure before relying on it.

• Current market numbers you supply in the chat (turnover, PCR, margin balance, northbound flows).
• An agent that supports skills folders or the Claude apps Skills setting.
• Access to a paid data source only if you want to use the optional indicators the file names.

Safety notes: scripts and network

The skill contains no scripts and makes no network calls, so it cannot pull prices or send your data anywhere on its own. Anything the agent says about current values comes from what you paste into the conversation or from other tools you have enabled.

The file also warns about data timing. Margin financing data is published with a one-day lag (T+1), and northbound disclosure rules changed after 2023, so real-time figures are less transparent than before.

• No scripts are included, so there is nothing to execute.
• No network access is used by the skill itself.
• Verify numbers against the exchange or data provider before acting on them.

Tips for better results

Give the agent dated, sourced numbers and state the market. The thresholds differ between A-shares, US stocks and crypto, so a single rule applied everywhere will mislead.

Treat the output as a checklist, not a trading decision. The skill itself notes that sentiment indicators are not precise timing tools and that extreme readings can persist for long periods.

• Name the market (A-shares, US, or crypto) in every request.
• Prefer weekly or monthly cumulative northbound flows over single-day figures, as the file suggests.
• Cross-check margin changes with ETF creation and redemption data.
• Filter out bots and marketing accounts before reading social-media sentiment.