Sovereign multi-tenant AI routing and RAG orchestration at the edge.
| Branch / track | Supported |
|---|---|
main |
Yes |
Do not open public issues for security flaws.
Report via private GitHub Security Advisory or the Chief Architect. Include tenant-isolation impact if relevant.
| Channel | Response |
|---|---|
| Dependabot (pip + Actions) | Merge security PRs promptly; pin floors for langsmith, pydantic-settings, Core |
| Code scanning / Bandit / Gitleaks | Fix on main; no secrets in tree |
| Red team / stress suites | Expand guardrails; never weaken tenant ACLs for convenience |
| Core SDK advisories | Bump coastal-alpine-core and re-run pytest |
| ID | Mitigation |
|---|---|
| GHSA-f4xh-w4cj-qxq8 | langsmith>=0.8.18 |
| GHSA-4xgf-cpjx-pc3j | pydantic-settings>=2.14.2 |
| Prompt injection | Core SecurityGuard on all inbound LLM messages |
| GITHUB_TOKEN sprawl | CI permissions: contents: read |
.env*,*.pem,*.key,secrets/gitignored.- Prefer mTLS MQTT (
ssl://…:8883) in production. - Pre-commit secret detection where enabled.
- CI: install Core + requirements, flake8, pytest.
- SecOps + red-team scheduled workflows.
- No silent exfiltration of personal or tenant operational data
- Prefer local-first processing; third-party AI only with explicit operator configuration and UI/docs disclosure
- Report vulnerabilities via GitHub Security Advisories or the maintainer contact on the org profile
- High-stakes production changes require human approval (HITL)
- We do not sell personal information or customer operational data to third parties.
- Optional AI or cloud services run only when configured by the operator; processing must be disclosed (in-product and/or docs).
- Prefer local-first paths so third-party transfer is unnecessary by default.
- Design in accordance with the Privacy Act 2020.
- Operate in accordance with Te Mana Raraunga principles for Māori data sovereignty interests.
- Align AI features with NZ AI safety / responsible AI expectations (HITL, transparency, no silent training on private content).