If you discover a security vulnerability in any package in this repository (@kernel.chat/kbot, @kernel.chat/kbot-finance, or the kernel.chat web application), please do not open a public issue.

Email isaacsight@gmail.com with the subject line:

security: <package> — <one-line summary>

Include:

• A description of the vulnerability and the conditions required to reproduce it.
• The package name and version where you observed it.
• Steps to reproduce (commands, request payloads, code snippets).
• Your assessment of the impact (e.g., remote code execution, data disclosure, audit log tampering).
• Any proof-of-concept code or screenshots that help confirm the issue.

You can expect:

• Acknowledgement within 48 hours.
• An initial triage response within 7 days with a rough timeline.
• A coordinated disclosure window — typically 30-90 days depending on severity. We'll work with you on the timing.
• Public credit in the release notes of the fix (unless you prefer to remain anonymous).

The following are in scope for vulnerability reports:

• @kernel.chat/kbot — the AI agent, its tool surface, BYOK key handling, MCP server adapters, computer-use approval gates.
• @kernel.chat/kbot-finance — content-addressed envelopes, the hash-chained audit log, regulatory verifier rules, the MCP server, engine adapters (Polymarket, EDGAR), approval token signing.
• kernel.chat web application — the magazine site, Supabase edge functions, auth flows, share modal, ChatGPT/Claude conversation import.

The following are out of scope:

• Third-party services we integrate with (Anthropic API, OpenAI API, Ollama, Polymarket Gamma API, SEC EDGAR). Report those upstream.
• Issues that require a malicious npm package installed alongside ours — npm-ecosystem-level threats belong to npm.
• Reports against forks or vendored copies of this code.

@kernel.chat/kbot-finance ships as an Apache 2.0 reference implementation. It has not been formally audited for production trading. If you find an issue specifically in the audit-grade primitives (hash chain integrity, envelope content-addressing, approval token verification), please flag the severity as high — a vulnerability in those primitives undermines the substrate's whole premise.