Engram stores an agent's episodic, semantic, and procedural memory, so its own trust boundaries matter: a corrupted or poisoned memory store is a credible path to steering an agent's future behavior. This document covers how to report a vulnerability.

Please report security issues privately, not in public issues or PRs:

• Open a GitHub private security advisory: https://github.com/TAIPANBOX/engram/security/advisories/new

Include the affected version/commit, a description, and a minimal reproduction. We aim to acknowledge within a few days and to fix high-severity issues before any public disclosure. There is no bug-bounty program; we credit reporters in the advisory unless you prefer otherwise.

Engram is pre-1.0; only main is supported. Fixes land on main and are not backported.

Every change must pass the full gate before merge: pytest, ruff check ., ruff format --check ., and mypy engram (strict). See CONTRIBUTING.md.