Following up on the Nemotron attestation example posted here recently ( GRC_Claw/examples/6g-nemotron-attestation at main · AAH20/GRC_Claw · GitHub ), I want to open this up properly rather than keep building it solo. There’s a real, specific opportunity for this community to make agent governance and post-quantum attestation a first-class part of the AI-RAN stack, not an afterthought bolted on once something goes wrong in production.
Where the project actually stands right now, honestly: the ML-KEM/ML-DSA signing path is real, audited-library-backed, and round-trip tested. The native, from-scratch lattice implementation is explicitly not production-ready yet, that’s documented plainly in the repo, not hidden. That gap is actually the best entry point for anyone who wants real, meaningful open source contribution rather than a cosmetic PR.
Concrete things worth collaborating on, if anyone here wants in:
- Deeper NemoClaw/OpenShell integration points. I built this deliberately complementary to NVIDIA’s own guardrail runtime, but I don’t have visibility into where it should actually hook into that flow in a real deployment. Anyone closer to that stack, this is the highest-value thing to get right together.
- Real NTT/lattice math for the native backend. Currently an honestly-labeled scaffold. Turning it into an independent, from-scratch RLWE implementation, reviewed properly, would be a serious contribution, not a toy PR.
- Testing against the Nemotron Large Telco Model specifically, the current example uses a general Nemotron model over NIM; pointing the same governance layer at the LTM and documenting what changes would be genuinely useful to everyone building on it.
- More attack scenarios in the policy firewall. The current policy check is intentionally simple (a blast-radius threshold). Real RAN-specific policy logic from people who actually operate this infrastructure would make this immediately more credible.
Why this is worth a community effort rather than staying one person’s side project: CNSA 2.0 sets a real 2027–2033 compliance timeline, and telecom infrastructure has a multi-decade operational life. Whatever governance layer becomes standard here will still be running, and still need to be trusted, long after the people who built it have moved on to something else. That’s worth getting right collectively, not owning individually.
Repo, issues open, contributions welcome: GitHub - AAH20/GRC_Claw: "The OSS chassis for ISO 42001-compliant agentic AI Swarm Harness, Anti-Swarm WAF, MAVLink & UAS compliance, a2zsoc.com bridge" · GitHub