Market Data Agent Skill

At a glance

The public skill name is Cornerstones Market Data Skill. The slug is cornerstones-market-data-context. The hosted skill file is the source an agent should read.

Use it when a human wants to give Claude Code, Codex, Hermes, OpenCLI, OpenClaw, Cursor, or another shell-capable agent a safe operating procedure for Cornerstones market data. The human copies one bootstrap prompt; the agent decides whether to create a persistent skill or follow the hosted file for the current task.

Copy this to your agent

Use the Cornerstones Market Data Skill at https://www.usecornerstones.com/skills/cornerstones-market-data-context/SKILL.md. If your agent runtime supports persistent or local skills, install or create a skill named `cornerstones-market-data-context` from that file. If it does not, read the file and follow it as the operating procedure for this task. Then install or update `cornerstones-client`, run `cornerstones-client --help`, `cornerstones-client guide`, and `cornerstones-client changelog`, verify my Cornerstones API key when I provide one, and check source labels, freshness, data quality, and account access before using any market data. Do not place trades or request broker/exchange credentials.

How it works across agents

• Codex / Claude Code: fetch the raw SKILL.md, then create or follow a local skill depending on the runtime's skill support.
• Hermes / OpenCLI / OpenClaw: read the same SKILL.md as the task procedure, then run the public client commands from the shell.
• Fallback path: if no persistent skill system exists, the agent still has everything it needs after reading the file.

The user should not need to choose GitHub, PyPI, or a marketplace. The skill is hosted as a public SKILL.md; the client package itself is installed from PyPI as cornerstones-client.

What the skill teaches

• Install and update cornerstones-client.
• Read live help before assuming a command exists.
• Start a browser trial for discovery, or store a Cornerstones API key with auth login and run verify.
• Keep Cornerstones keys separate from broker, exchange, model, or upstream data-provider keys.
• Use only the market surfaces available to the provided Cornerstones key.
• If a surface is unavailable, choose an allowed read instead of guessing.
• Use Cornerstones as read-only market context before action; it does not place trades.

First commands

python -m pip install -U cornerstones-client
cornerstones-client --help
cornerstones-client guide
cornerstones-client changelog
cornerstones-client auth status

Start a browser trial and prove discovery:

cornerstones-client trial start
cornerstones-client trial token
cornerstones-client guide
cornerstones-client changelog

After the dashboard issues a durable key:

cornerstones-client auth login --api-key <issued-api-key>
cornerstones-client verify
cornerstones-client guide

Market data checks

cornerstones-client fx quote --symbol EURUSD
cornerstones-client fx bars --symbol XAUUSD --timeframe 1h --count 3
cornerstones-client stocks quote --symbol AAPL
cornerstones-client stocks quote --symbol 600519.SS
cornerstones-client chart fx --symbol XAUUSD --timeframe 15m --bars 200 --indicator ema20 --indicator rsi14
cornerstones-client events recent
cornerstones-client orderflow summary --symbol XAUUSD

Charts, context, options, macro exact series, and event export require Pro or Max. Orderflow raw, summary, context, historical reads, and liquidity metrics require Max. If a surface is unavailable, choose a permitted read instead.

Reading responses

Before using any market data, check:

• source labels
• freshness or timestamp fields
• data quality notes
• account access state
• source and symbol normalization fields when present

Skill links

• Open the hosted skill file
• Open the copy prompt
• Public path after deploy: /skills/cornerstones-market-data-context/SKILL.md