Skip to content

Security: ellmos-ai/bundles

Security

SECURITY.md

Security Policy — bundles

Overview

bundles provides the declarative recipe layer of the ellmos ecosystem: bundle manifests, catalogues, and composition knowledge. Security in bundles centers around static declarative integrity, cryptographic hash verification, zero external network egress, and non-elevated tooling.

Security Architecture & Principles

1. Pure Declarative & Static Data

  • bundles consists exclusively of structured JSON manifests, schemas, contracts, documentation, and frozen visual design tokens.
  • There are no executable runtime binaries, no background daemon processes, and no dynamic code evaluation pipelines.

2. Local-First & Zero Network Egress

  • All tools in tools/ (tools/generate_banner.py, tools/export_from_source.py) and test suites operate entirely offline.
  • The repository requires zero network dependencies, opens no network ports, and transmits no telemetry or background requests.

3. Bit-Identical Reproducibility & Cryptographic Hashes

  • Every bundle manifest, contract, catalog, and export receipt carries an explicit canonical SHA-256 content_hash.
  • Integrity is continuously validated via automated contract test suites (tests/test_bundle_contracts.py), preventing tampering or bit rot.

4. Non-Elevated Execution & Privilege Isolation

  • Repository tools and automation scripts run in standard user mode (RunAsInvoker).
  • No administrative or elevated operating system privileges are required or requested.

Supported Versions

Version Supported
1.0.x ✅
< 1.0.0 ❌

Reporting a Vulnerability

If you discover a security vulnerability or integrity hazard in bundles:

  1. Do not open a public issue.
  2. Report the vulnerability privately via GitHub Security Advisories at https://github.com/ellmos-ai/bundles/security/advisories or directly to the repository maintainers.
  3. Include detailed steps to reproduce, affected manifests or tools, and observed behavior.
  4. Reports are acknowledged within 48 hours, and resolutions will be coordinated prior to publication.

There aren't any published security advisories