Security policy

Supported versions

Security fixes are applied to the default branch (main) only. Use the latest commit for production-like deployments.

Reporting a vulnerability

Please do not open a public GitHub issue for security vulnerabilities.

Instead:

1. Open a private security advisory on GitHub (Repository → Security → Advisories → Report a vulnerability), or
2. Contact maintainers through a private channel if one is listed on the repository.

Include:

• Description of the issue and impact
• Steps to reproduce (if possible)
• Suggested fix (if you have one)

We aim to acknowledge reports within a few days.

Secrets and accidental exposure

• Never commit real API keys, tokens, or .env files. Use .env.example only with placeholders.
• If you accidentally committed a secret:
• Rotate/revoke the key immediately with the provider (OpenAI, Serper, etc.).
• Remove the secret from the latest commit and open a PR.
• If the secret was pushed to a public remote, assume it is compromised even after removal from history.

GitHub secret scanning may alert on known patterns; treat alerts seriously.

Scope and threat model

StockSage is a single-user, local-only tool. It has no authentication, no session management, and no multi-tenant isolation. The FastAPI server binds to 127.0.0.1:8000 by default — do not expose it on a public interface without adding auth in front (e.g. a reverse proxy with basic auth). If you use --host 0.0.0.0 or deploy it on a VPS, anyone who can reach that port can trigger LLM calls using your API keys.

Primary threat surface: - Dependency vulnerabilities — pip-audit or uv audit can scan the lock file. - SSRF via misconfigured tools — the search tool makes outbound HTTP; Ollama base URL is configurable. - Accidental secret exposure — API keys in .env, logs, or issue reports. - Malicious ticker input — symbol is validated against a format regex before any I/O, but downstream CSV filenames are derived from the symbol; verify this if you extend the storage layer.