Skip to content

SuarezPM/Apohara-Guard

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Apohara Guard

Isolation primitives + ML-subprocess sandbox patterns for the Apohara PROBANT ecosystem.

Part of the cross-AI code verification trinity submitted to TechEx 2026:

Why AGPL-3.0?

The other three repos ship under Apache-2.0 to lower friction for enterprise adoption of safety tooling. Apohara Guard is specifically AGPL-3.0 so that downstream improvements to the sandbox + scanner detection patterns flow back to the public commons.

Status

Hackathon-submission snapshot, 2026-05-18.

Tracked code surface in this repo (intentionally minimal — see git ls-files):

  • src/sandbox/index.ts — bwrap mount-ns + user-ns + RLIMIT_AS sandbox primitives (3-layer active)
  • src/scanner/index.ts — content scanner stub
  • tests/sandbox.test.ts — sandbox primitives test suite
  • docs/research/sandbox-design.md — design rationale
  • THIRD_PARTY_NOTICES.md — attribution to RAPTOR + bubblewrap

Honest disclosures (per Apohara AUDIT.md culture):

  • Sandbox is 3 active + 2 planned, not 5 active. The 5-layer kernel sandbox naming was corrected to honest framing in commit ed1539c. Landlock LSM + seccomp-bpf require libseccomp bpf-blob generation and are scheduled post-hackathon.
  • Build tooling (package.json, bun.lock, Dockerfile, biome.json) is not yet tracked. Full local-dev quickstart lands post-hackathon. For now: clone + read the 6 tracked files + cross-reference the design doc.
  • Content scanner is a stub. Real detection logic, training data attribution, and threat-model boundaries land post-hackathon.

Cross-references

License

GNU Affero General Public License v3.0 only — AGPL-3.0-only.

Contact

About

Isolation primitives + ML-subprocess sandbox patterns for the Apohara PROBANT ecosystem. AGPL-3.0.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors