Observed arrival · 2026-11-15

Oberig puts a local checkpoint between coding agents and their inputs

oberiglabs.com Visit website

Editorial interest 78/100 Selection signal · not a rating of the site

A local guard for Claude Code, Codex, and OpenCode that says it verifies its AI endpoint, masks secrets, and filters hidden instructions before they reach the model.

Landing page captured for the 2026-11-15 issue.

For: Claude Code, Codex, and OpenCode users
Worth noticing: The project reports catching 92.8% of 600 hidden injections in 22 languages, with a 1.4% false-positive rate across 222 real documents.

Field notes

The displayed verification sequence uses a fresh nonce, rechecks attestation every 10 minutes, and checks response signatures against the exact sent and received bytes. The page also describes scrubbed copies of Claude Code and Codex traces, with secrets, personal data, paths, usernames, and hostnames masked while originals remain unchanged. Listed setup requirements are Node 18.17 or newer and a NEAR AI Cloud key.

Observed signals

Read the marks

Editorial observations of this landing page, not a rating.

○OpenPublic substance visible

✦PrettyNotable craft visible

●ProPolished or operationally mature

◎NicheUnusually specific use

One card from the complete issue

Coffee on the Tukwila Sidewalk

282,335 arrived 1,000 judged 985 catalogued Enter the complete issue