skills.rest
122yjs
122yjsCOMMUNITY
Explore all skills by this author →
GitHub

invest-harness

Generates multi-agent equity research reports with evidence planning, QA, and workspace handoffs.

Updated May 9, 2026
Search Tags:#multi-agent#mcp#financial analysis#stock analysis#valuation#equity research#investment report
One-click install
npx skills add https://github.com/122yjs/invest_harness --skill invest-harness-122yjs
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: invest-harness
Source: https://github.com/122yjs/invest_harness
Command: npx skills add https://github.com/122yjs/invest_harness --skill invest-harness-122yjs

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Producing a thorough, source-cited investment research report on a listed company normally requires coordinating financial, qualitative, valuation, technical, macro, and risk analysis by hand. This Skill decomposes that work into specialized analyst roles, runs them through a reproducible file-based workspace, and assembles a QA-reviewed 18-section final report. ## Core Features & Use Cases - Multi-agent research pipeline: Orchestrates financial, fundamental, valuation, technical, macro-sentiment, and risk-scenario analysts, then synthesizes and QA-reviews their findings. - Evidence and source governance: Plans evidence before analysis, prioritizes T0 official disclosures (SEC EDGAR, DART/KRX, company IR), and records data gaps and source conflicts instead of guessing. - Reproducible workspace contract: Stores every intermediate artifact under a dynamic ${ACTIVE_WORKSPACE}/ directory with stage gates, evidence ledgers, and claim-evidence maps. - Use Case: Ask for a deep-dive report on Apple (AAPL) with a mixed focus and 5 peer companies; the harness auto-identifies the company, collects data via MCP servers or official APIs, and delivers a rated report with price target scenarios. ## Quick Start Ask the agent to follow AGENTS.md and docs/harness/invest/runbook.md and use the invest-orchestrator skill to analyze a company such as Apple (AAPL, NASDAQ) with standard depth and a mixed investor perspective.

Frequently Asked Questions about invest-harness

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate an investment research report for a stock?▼

Provide a company name or ticker and optional preferences for focus, horizon, peer count, and depth. The orchestrator auto-identifies the company, runs specialized analyst roles, and assembles an 18-section report with rating and price target scenarios.

What data sources does the investment research workflow use?▼

It prioritizes T0 official disclosures such as SEC EDGAR, DART/KRX, and company IR filings. Vendor snapshots like yfinance, FMP, and Alpha Vantage serve as cross-checks, and web search plus fetch acts as a fallback when APIs are unavailable.

Does the harness support Korean listed companies?▼

Yes, Korean companies are supported through the korea-stock MCP server, which calls DART and KRX official APIs for disclosures, XBRL financial statements, and daily price data. yfinance is used as a secondary cross-check for Korean tickers.

Can I use this with Claude, Codex, or other AI agents?▼

Yes, the harness is runtime-agnostic and defined entirely in Markdown contracts. Entry points exist for Claude, Gemini, Hermes, OpenClaw, Codex, and opencode, and role SKILL.md files can be referenced directly in prompts.

What happens when data sources conflict or are unavailable?▼

Conflicting figures are recorded in a conflicts file with differences and causes rather than averaged. Unavailable sources are logged as data gaps in the source call plan, and missing items are marked as unverified or requiring further confirmation.

Is the generated report personalized investment advice?▼

No, the output is explicitly informational analysis, not personalized investment advice or trade solicitation. Every report states this boundary, and QA review checks for overconfident claims and unsourced assertions.

Related Skills

Explore More Finance & Accounting Skills→
tjdwls101010tjdwls101010

marketdata

Retrieve multi-source financial data from YFinance, FRED, SEC EDGAR, Finviz, CFTC, CBOE, and YCharts.

Community
Advanced
gganbukim1gganbukim1

data-scraper-agent

Scrape public web sources on a schedule and enrich items with Gemini summaries.

Community
Advanced
anavvanzinanavvanzin

agency-investment-researcher

Produce institutional-quality investment research with primary-source citations and scenario modeling.

Community
Advanced
Bitterbot-AIBitterbot-AI
2.5k

finance-data

Fetch real-time stock quotes and market data from Yahoo Finance endpoints.

Official
Intermediate