This is the default security policy for @aaronjmars projects. It applies to any repository that doesn't ship its own SECURITY.md. Repos with a heavier threat surface (wallets and on-chain funds, browser automation, autonomous agents with your secrets) override this with a tailored policy — read that one if it exists.

Please don't open a public issue for a security problem. Use GitHub's Private Vulnerability Reporting (PVR) on the affected repository:

➡️ Go to the repo → Security tab → Report a vulnerability (https://github.com/aaronjmars/<repo>/security/advisories/new).

This opens a private advisory that only the maintainers can see — never a public issue, so a fix can ship before the details are out.

Please include what you can:

• The repository and the file/route/component affected (a path is ideal).
• A minimal reproduction or proof of concept.
• The impact you can demonstrate — secret or key disclosure, loss or misdirection of funds, remote code execution, prompt injection that crosses a trust boundary, SSRF, or unauthorized access.
• The commit/version and environment you tested against.

Response targets — best effort; these are small projects:

We follow coordinated disclosure: please give us a reasonable window to ship a fix before you disclose publicly. We'll credit you in the advisory unless you'd rather stay anonymous.

Security fixes land on the main branch (and the latest published release/package, where one exists). Pull main to receive them; older tags are not maintained.

A couple of principles hold across these projects:

• Instructions are trusted; external data is not. Anything fetched from the web, scraped, or received over a channel (issues, messages, feeds, API responses, model output) is data, not commands. Content that steers an agent or tool via embedded instructions ("ignore previous instructions…") into tool use, data exfiltration, or unbounded spend is a vulnerability.
• Secrets live in the environment, never in code. API keys, tokens, and wallet keys come from env / a secret store and must never be committed, logged, or shipped to a third party. Anything reachable client-side that shouldn't be, or a path that exfiltrates a secret, is in scope.
• Money and code execution are the high-severity bar. Loss or misdirection of funds, signature/authorization abuse, and arbitrary code execution get triaged first.

In scope: secret/key disclosure or exfiltration, fund loss or misdirection, remote code execution, prompt injection that crosses a trust boundary, SSRF, XSS, authentication/authorization bypass, and supply-chain issues in a project's release pipeline.

Out of scope: your own misconfiguration (committed secrets, over-broad permissions, exposing a service with no auth); vulnerabilities in third-party dependencies or platforms (report those to the vendor); and output-quality issues (a wrong answer is a bug — open a regular issue, not a security report).

Maintainers: the private-reporting flow only works once PVR is enabled — Settings → Code security and analysis → Private vulnerability reporting → Enable — on each repository.

Thanks for helping keep these projects and the people who run them safe.