AlphaGBM Skills
Real-data options & stock intelligence skills for AI agents — 29 skills with built-in demo data, no API key required to try.
- Source repo
- AlphaGBM/skills
- Stars
- ★ 3.6k
- Last updated
- 24d ago
- License
- MIT
- Primary language
- Python
- FA score
- 43/100 · Major gaps
At a glance
- How it runs
- Works with
- Platform-specificClaude CodeCodex (Partial support)
- Cost
- Free tier plus a paid hosted plan
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- A trader asks 'Analyze AAPL' to get a comprehensive stock score and risk assessment. (Works with demo data.)
- Not a fit if
- Teams that need fully offline research without a vendor cloud API
- Users needing an MCP server or non-skills-protocol integration
- Researchers needing EU, JP or KR options coverage
- Source review
- 43/100 · Major gaps 2 safety controls not found
What does this agent do, and when should you use it?
AlphaGBM Skills is a set of 29 AI skills for options and stock research, delivering real market data instead of LLM guesses. It caters to AI agents like Claude Code and Cursor, and includes a CLI tool. Skills cover core analysis (stock analysis, options scoring, strategy building), data intelligence (IV rank, unusual activity), risk & portfolio discipline (hedge advisor, BPS backtest), and investment master methodologies. Each skill ships with mock data (e.g., AAPL, NVDA) for offline testing, while live data requires an API key. The repository provides clear installation instructions and usage examples.
The repository provides 29 skills, each implemented as a directory with instructions and optional code. Skills make HTTP calls to the AlphaGBM API (hosted at alphagbm.zeabur.app) or use built-in mock data. For example, 'Stock Analysis' computes a G=B+M score (basics + momentum) and returns a 0-10 score, EV, and risk. 'Options Score' returns scores 0-100 for four strategies. Other skills include IV Rank, Vol Surface, Greeks, P&L Simulator, and more. The CLI (alphagbm) wraps these functions, e.g., 'alphagbm stock analyze AAPL'.
- A trader asks 'Analyze AAPL' to get a comprehensive stock score and risk assessment. (Works with demo data.)
- An options trader asks 'Score NVDA options' to compare strategies and pick a direction.
- An investor asks 'Add NVDA to my knowledge base' to automatically create research files and track triggers.
- A portfolio manager asks 'Hedge my AAPL at cost 140, now 180' to get scenario-based hedge recommendations.
- A quant backtests 'BPS on QQQ' to see if FearScore signals improve returns.
- A trader checks 'Is TSLA IV high?' using IV Rank and VRP metrics.
How do you install or deploy this agent?
- Clone the skills repo into your project: 'git clone https://github.com/AlphaGBM/skills.git .claude/skills/alphagbm' (or .cursor/skills/alphagbm for Cursor).
- For the CLI: 'cd skills/cli && pip install -e .'.
- Set your API key: 'alphagbm config set-key agbm_xxxxxxxxxxxxxxxx' (or export ALPHAGBM_API_KEY).
How do you use this agent?
- After installation, prompt your AI agent, e.g., 'Analyze AAPL stock using AlphaGBM' — demo data works immediately.
- For live data, set ALPHAGBM_API_KEY (free at alphagbm.com/api-keys) and optionally ALPHAGBM_BASE_URL.
- Check API health with 'curl https://alphagbm.zeabur.app/api/health'.
- Use the CLI: 'alphagbm stock analyze AAPL', 'alphagbm options score NVDA'.
What are this agent's strengths and limitations?
- Real market data with verifiable computations (e.g., IV via Black-Scholes, VRP as IV-HV).
- 29 skills covering analysis, strategy, risk, and master methodologies.
- Built-in demo data allows full functionality without an API key.
- Multi-market coverage (US, HK, CN, commodities).
- CLI and API offer programmatic access beyond natural language.
- Transparent scoring models (e.g., G=B+M, option scoring formula).
- Live data requires an API key and has strict rate limits on free tier (2 stock analyses/day, 1 option analysis/day).
- Dependent on AlphaGBM's cloud API; if service changes, skills may break.
- Limited market coverage (no EU, JP, KR options yet).
- Skills are tailored for agents that support the skills protocol (Claude Code, Cursor); other agents may need adaptation.
- No MCP server provided; integration is via file-based skills.
How does this agent compare with similar options?
Compared to LLM roleplay tools that generate 'fake' data, AlphaGBM provides real, verifiable options data. Compared to generic finance APIs, AlphaGBM offers deeper options analytics (IV/HV/VRP/Greeks/Skew) and quantitative scoring.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| AlphaGBM Skills This agent | 43 · Major gaps | Agent plugin / skillFreemium | ★ 3.6k | 24d ago | Python | Claude Code |
| Vibe-Trading | 64 · Some gaps | CLIFree + model costs | ★ 36k | 1d ago | Python | OpenAI API · Claude API |
| QuantDinger AI Trading OS | 89 · Good | Self-hosted serviceFree + model costs | ★ 13k | today | Python | Codex · Claude Code · OpenAI API |
| FinRobot | 51 · Major gaps | CLIFree + model costs | ★ 8.2k | 20d ago | Python | OpenAI API |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows: README describes API key setup (ALPHAGBM_API_KEY) and external API calls, but does not explain least privilege or data flow details; there are signs of user confirmation (e.g., 'Ask your AI' interaction), but no explicit requirement; data flow transparency is limited, only mentioning API calls; sensitive data handling is not mentioned; dependency security is not mentioned; external effects include API calls and possible file writes (e.g., knowledge base), but side effects are not described; rollback is not mentioned; source attribution is partially present (e.g., 'Every number has a source'), but specific sources are not provided. Deductions: lack of concrete evidence for least privilege, user confirmation, data flow transparency, sensitive data handling, dependency security, external effects, and rollback.
Evidence shows: skill list and descriptions are consistent, architecture diagram is clear, but no tests or error handling details are provided; dependency availability is not explicit, but built-in demo data is mentioned; failure messages are not mentioned. Deductions: lack of error handling documentation and dependency availability guarantees.
Evidence shows: clearly targets traders and AI agents, provides multiple use cases and examples; capability boundaries are explained through skill list and architecture; trigger precision is partial, example queries are specific, but no precise trigger rules; environment fit is good, supports Claude Code, Cursor, etc. Deductions: trigger precision lacks precise rules.
Evidence shows: information architecture is clear, with skill categories and architecture diagram; installation instructions are detailed, including multiple methods; naming is stable, skill names are consistent; examples and FAQ are provided (e.g., example workflow); known limitations are partially mentioned (e.g., 'TODO' and roadmap), but not comprehensive; license is MIT, clear; versioning changelog is missing; maintenance responsibility is not explicit. Deductions: missing versioning changelog, maintenance responsibility unclear.
Evidence shows: output usability is high, with specific output examples (e.g., scores, target prices); marginal value is high, with unique features (e.g., FearScore, BPS Backtest); cost-benefit is reasonable, with free tier and demo data. Deductions: no evidence of actual performance or user feedback.
Evidence shows: claims every number has a source, but specific sources are not provided; cross-source corroboration is insufficient, relying only on own API; fact-inference separation is partial, e.g., 'G=B+M' model, but not clearly distinguished. Deductions: lack of specific sources and cross-source verification.
- Not found in source: dependency securityPin versions and run a dependency audit (npm audit, pip-audit) before installing; prefer running it in a container.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- The repository provides no tests or verification evidence; all functionality descriptions are based on README claims, and actual runtime behavior is unknown.
- API keys and external API calls pose security risks, but no security best practices or data flow explanations are provided.
- No versioning changelog is provided, and maintenance responsibility is unclear, which may affect long-term reliability.