feedBulletin is pre-1.0. Only the latest released version receives security fixes. Older versions are not patched.

Do not file a public GitHub issue for a security report.

Email ope@312.dev with:

• A description of the vulnerability and its impact.
• Steps to reproduce, ideally with a minimal feeds.yaml or a sample of the malicious input.
• Your assessment of severity (low / medium / high / critical) and any CVSS estimate you'd like to suggest.
• Whether you'd like to be credited publicly in the fix changelog.

We acknowledge reports within 72 hours and aim to ship a fix within 14 days for medium/high severity issues. Critical issues get expedited handling.

The interesting attack surface in a forum aggregator:

• HTML injection through scraped post bodies — we sanitize via ammonia, but bugs are possible. Any HTML escape from a learned site profile that lands rendered in the WebView is in scope.
• YAML deserialization — feeds.yaml is user-controlled but read locally; a malicious YAML file that causes the app to write outside its app-data directory, exfiltrate the API key, or escalate to RCE is in scope.
• Anthropic API key exposure — the key is read from ANTHROPIC_API_KEY at runtime. Any code path that logs it, ships it to a third party, or persists it to disk in plaintext is in scope.
• Cookie cache leakage — the cookie cache stores session tokens for forums that gate behind login. Cross-site reads, dumping the cache to logs, etc. are in scope.
• Selector-learner prompt injection — a forum that serves malicious HTML designed to manipulate the Anthropic selector-learner into producing harmful selectors (e.g. ones that exfiltrate page content) is in scope.

• The Anthropic API itself. Report API issues to Anthropic.
• The forum sites we scrape. We treat them as untrusted but we don't own their security.
• Denial-of-service from misconfigured polling (set a reasonable poll_interval_s and don't bury your machine).
• Issues that require physical access to an already-unlocked machine.

We follow coordinated disclosure: we fix, ship, then publish details in the release notes. Embargo extensions are negotiable for severe issues.

Reporters who prefer not to be named will not be named. Reporters who want credit get a line in the release notes and the optional CVE assignment.

To everyone who takes the time to look at the code and report responsibly.