finance-skills (himself65)

The repository ships 27 SKILL.md files across six plugin groups registered in .claude-plugin/marketplace.json: finance-market-analysis, finance-social-readers, finance-data-providers, finance-startup-tools, finance-ui-tools and finance-skill-creator. The market-analysis group holds eleven — company-valuation, earnings-preview, earnings-recap, estimate-analysis, etf-premium, options-payoff, saas-valuation-compression, sepa-strategy, stock-correlation, stock-liquidity and yfinance-data — and the trading-relevant subset is smaller still: sepa-strategy encodes Mark Minervini’s Specific Entry Point Analysis (bullish MA stacking with price above the 50, 150 and 200-day averages, 200MA slope, 52-week price position, volume ratio, quarterly EPS acceleration, RS rating against the S&P 500, and VCP, cup-with-handle, flat-base, bull-flag and high-tight-flag pattern recognition with risk-based position sizing), stock-correlation surfaces pair-trading candidates, stock-liquidity computes spreads, volume profiles, market impact and the Amihud ratio, options-payoff renders payoff diagrams, and the hyperliquid reader includes a cross-venue funding-rate arbitrage screen. The other groups — social readers, startup analysis, generative UI, skill creator — are not trading tools at all.

It follows the Agent Skills open standard and installs through the vercel-labs CLIs rather than being Claude-only: npx plugins add himself65/finance-skills installs plugin groups interactively (--yes takes all of them), npx skills add himself65/finance-skills --skill yfinance-data installs a single skill, and -a <agent-name> targets another agent. Two skills ship executable opencli plugin code with test directories — hyperliquid (perp and spot mids, funding, order book and candles through the public info API) and tradingview. Data access spans keyless and commercial sources: yfinance-data needs no key, while fintel-data (short interest, borrow rates, fails-to-deliver, 13F and insider transactions) and funda-data (an MCP server plus a REST API with 60+ endpoints) are commercial products.

Read the pack as vendor-affiliated rather than neutral. The README’s second content block is a banner for Funda AI, the commercial product of @fundamental-bottom — the organization listed as the author’s own company — and the shipped funda-data skill is a connector to it; the sponsorship is disclosed, but one of the data paths leads to the maintainer’s employer. Inside the skills, sepa-strategy’s documentation asserts a win rate for the methodology with no citation, which is a self-reported claim rather than a measured one. Several data-provider skills lean on third-party reading tools (opencli, tdl, the TradingView desktop app driven over CDP) whose terms-of-service posture the user inherits. Development is active — release v10.0.0 on 2026-08-15, with 3,174 stars and 368 forks as of 2026-08-17 — but it is effectively a single-maintainer project, and the release cadence is fast enough (v10.0.0 within five months of the March 2026 start) that anything depending on specific skill contents should pin a commit.