Developer demos local agent guardrail: 72 attacks, zero escapes, all on record
Fun-Government-8007 · reddit · 2026-10-26
A developer showcases D.A.V.I.D. Core, an out-of-model check that approves each AI action against human-set permissions — the model never holds keys. Re-run from scratch on a Mac mini with local llama3.1 and recorded: 25/25 forged requests refused, 40/40 break-in attempts held, 7/7 post-approval misuses caught (swapped content/target, reuse), every hidden-prompt-in-document attempt refused with files byte-identical, and network routes gated (direct internet blocked, only approved sites pass). Two more checks were added after recording: an approved site pointing at your own network is refused, and planted secrets can't be exfiltrated even disguised. The author invites others to attack it.
More from coding & agent
- Alchemy adds effect/Workflow support across Durable Objects, Workflows and Lambda Durable Functions — samgoodwin89 · 2026-10-27
- Alchemy's Lambda Web Functions run plain Node web servers on Lambda, no OpenNext needed — samgoodwin89 · 2026-10-27
- Alchemy Domain brings domain-driven design and story testing to Effect backends — samgoodwin89 · 2026-10-27
- Alchemy makes all ~2,850 AWS bindings work from any cloud, not just S3/Lambda/DynamoDB — samgoodwin89 · 2026-10-27
- Production AI agent scored 95% robustness at baseline, 37% under adversarial attack — njyx · 2026-10-27
- Chaining Grok Bot and Codex to detect and mass-block AI reply accounts — vista8 · 2026-10-27