benchmark: antfleet two-model review on ronkenx9/agentfloat-hook @kenn_ronin findings (unanimous claude + gpt-5): → [HIGH] watcher always uses testnet chain config, ignoring mainnet CONFIG.chainId → [HIGH] FloatVault withdraw ignores strategy actualOut, transfers requested amount → approve path m
@antfleetdev·1 jun 2026·2 fuentes
Leer artículoProyectos relacionados
Todas las fuentes
benchmark: antfleet two-model review on ronkenx9/agentfloat-hook @kenn_ronin findings (unanimous claude + gpt-5): → [HIGH] watcher always uses testnet chain config, ignoring mainnet CONFIG.chainId → [HIGH] FloatVault withdraw ignores strategy actualOut, transfers requested amount → approve path m
@antfleetdev
1 jun 2026
benchmark: antfleet two-model review on fathah/hermes-desktop @fathah_cr findings (unanimous opus 4.7 + gpt-5): → YAML writes quote values without escaping, corrupting config.yaml → external URL allowlist permits insecure http: scheme → setModelConfig api_key insertion assumes double-quoted YAML →
@antfleetdev
1 jun 2026
