Skip to content

Security: Ian-Tharp/Dynamic-Subgraphs

Security

SECURITY.md

Security Policy

Supported versions

Dynamic Subgraphs is pre-1.0. Security fixes are applied to the latest released 0.x version. Pin a version in production and upgrade promptly.

Reporting a vulnerability

Please do not open a public issue for security problems. Instead, report privately via GitHub Security Advisories ("Report a vulnerability" on the repository's Security tab), or email praht09ian@gmail.com.

Include reproduction steps and affected versions. We aim to acknowledge within a few days and will coordinate a fix and disclosure timeline with you.

Handling secrets

This SDK reads provider credentials from the environment (e.g. OPENAI_API_KEY, ANTHROPIC_API_KEY) and an optional TAVILY_API_KEY.

  • Keep secrets in a local .env (already git-ignored) or your platform's secret manager. Never commit real keys — see .env.example.
  • Local providers (LM Studio / Ollama) need no key; use them for fully offline runs.
  • Run recording (record=...) writes prompts, traces, and outputs to runs/<run_id>/. Treat that directory as potentially sensitive and keep it out of version control (it is git-ignored by default).

Safety model

By design the planner emits plans (a validated GraphSpec), not executable code. The compiler only instantiates registry-approved node kinds, and LangGraph types stay behind the compiler/runtime boundaries. There is no arbitrary code execution path from model output in v1. Tool-calling nodes are limited to the allowlisted tools in the registry. If you extend the registry with new node kinds or tools, you are responsible for their safety.

There aren't any published security advisories