Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
fix(audit-console): widen the environment-noise filter for the console sweep
nirholas07d662483d37
diff --git a/scripts/audit-console.mjs b/scripts/audit-console.mjsindex 42fee68c5..5fef64460 100644--- a/scripts/audit-console.mjs+++ b/scripts/audit-console.mjs@@ -13,9 +13,17 @@* Any route that fails in the parallel pass is re-run once, serially, and that* second reading is what the report carries. This worktree is shared with other* agents and a sweep routinely runs beside somebody else's full build; a page- * that misses its settle window at load 100+ has not failed. A real defect- * reproduces on the quiet retry, so nothing is excused, only re-measured, and- * the report says how many routes cleared that way.+ * that misses its settle window at load 100+ has not failed. The report says how+ * many routes cleared that way.+ *+ * The retry clears a momentary spike. It does NOT clear a sweep run while a+ * multi-minute build is going, because the retry runs under the same load, so a+ * surviving failure is evidence but not proof. On 2026-09-09 a sweep at load 80+ * to 100 reported 31 failing routes and 29 of them passed on a quiet box, all+ * with the same signature (a texture or module fetch dropped mid-load). Before+ * acting on a failure list: read the load, and if it is high, re-run just the+ * failures once the box is quiet, against a dev server you started yourself.+ * docs/audit/console-sweep-2026-09-09.md carries that comparison.** Usage:* node scripts/audit-console.mjs # every HTML route, both viewports