麦赛尔夫v1.0.4

A portfolio manager needs a dense, sourced report on a US-listed semiconductor or cloud company immediately after its latest quarterly release. The agent fetches the newest filing, calculates valuation multiples independently, and delivers the report with inline source links for every data point. Reports are saved in English-named files under RESEARCH/semiconductors/.

A family office is considering participating in a US IPO and needs the prospectus parsed into structured financials, comparable-company valuation and analyst forecasts for 2026E–2028E. The agent follows the IPO report template and returns a .md file sent via the messaging channel, without subjective recommendations.

A macro fund wants a weekly data-first update on a US sector (e.g., AI infrastructure) with commodity and stock prices no older than 12 months and explicit data dates. The agent stores raw structured data with source URLs in DATA/ and compiles a report in the principal's language, keeping file paths in full English.

An independent research boutique needs all output to comply with a strict citation format: every figure followed by an inline link such as (Amazon IR, Oct 2025). The agent enforces the non-negotiable rules in SKILLS.md, quotes and organizes external analyst views with attribution, and never inserts its own conclusions.

A Shanghai-based fund manager writes in Chinese but invests in US stocks, requiring report bodies in Chinese while keeping all file paths and filenames in full English. The agent acknowledges each task, fetches the latest financial period, and delivers a concise, information-dense .md report through the messaging channel.

The skill is installed once in an OpenClaw workspace by copying the asset templates (SOUL.md, AGENTS.md, SKILLS.md, USER.md, MEMORY.md, HEARTBEAT.md) and customizing USER.md for the principal. Value comes from replacing manual research workflows with a consistent, rules-bound research agent.

An operator deploys the agent across multiple principals or client mandates, delivering on-demand US stock deep dives and IPO reports as .md files. The strict citation and no-self-analysis rules make output auditable and defensible for professional investors.

The agent's DATA/ directory accumulates structured, sourced and timestamped US equity and commodity datasets. These structured outputs can be licensed or reused internally for further analytics, screening or downstream reporting.