Please report security issues privately to me@eismann.consulting — not via public GitHub issues. Include what you found, where (file/component), and how to reproduce it. I'm a solo maintainer: expect an acknowledgement within a few days and an honest assessment of severity and timeline; credit is given in the fix's release notes unless you prefer otherwise.

The project is in active pre-1.0 development; only the current main branch is supported. There are no backported fixes.

Reports I'm particularly interested in:

• Prompt-injection paths: a crafted requirements document, source schema, or profiling file steering the LLM agents into unsafe output. The deterministic validator gates and human checkpoint are the designed mitigations — findings that bypass both are high-severity.
• Generated-code issues: generated dbt/AutomateDV output that could execute unintended SQL against the target warehouse.
• Classic dependency or CI issues (the lockfile and workflows are in-repo).

One caution for reproduction material: run traces (.vault-agent/traces/) contain the full input documents and source metadata. Never attach traces from real data to a report — reproduce with the shipped synthetic examples under examples/inputs/.