Results for “Sub-agents”

Why it matters

Simon Willison tested Codex Desktop using GPT-5.6 Sol Ultra in aggressive sub-agent mode to construct a complete 3D web game from a single prompt. While Codex coordinated sub-agents and generated visual assets using gpt-image-2, it failed to fix visual rendering bugs during screenshot self-inspection.

Why it matters

Cloudflare released an open-source coding agent skill that transforms tools like Claude Code into automated security auditors. It runs a structured six-phase pipeline using adversarial validation and isolated sub-agents to trace and verify vulnerabilities.

Why it matters

Harness is a framework designed to automatically generate, configure, and orchestrate specialized teams of AI agents tailored to your project goals. It parses requirements to deploy focused sub-agents, minimizing generalist context bloat. Easily structure complex development tasks into isolated, collaborative roles.

Why it matters

Google introduced Gemini 3.6 Flash and Gemini 3.5 Flash-Lite, designed to lower latency and token usage in agentic workflows. Gemini 3.6 Flash reduces output tokens by 17% while costing $1.50/1M input and $7.50/1M output, and adds native client-side computer use capabilities.

Why it matters

Developer community feedback highlights recurring struggles with Codex in handling complex frontend design and sub-agent orchestration. Despite rapid iterations, core performance remains inconsistent for production-grade UI tasks.

Why it matters

Inspection of OpenAI Codex's core agent harness reveals a dedicated awaiter builtin configured via TOML and prompt instructions. It uses exponential backoff polling timeouts to await task completion while instructing the LLM to behave deterministically.

Why it matters

Amp has introduced Orbs, dedicated remote cloud environments designed to run AI coding agents asynchronously without eating local machine resources. Developers can launch multiple parallel agent threads via CLI flags and pull modified files back to local checkouts.

Why it matters

Sakana AI launched Fugu-Cyber (fugu-cyber-v1.0), a specialized cybersecurity endpoint for its Fugu agent orchestrator. The model scored 86.9% on the CyberGym vulnerability verification benchmark and 72.1% on Microsoft's CTI-REALM detection benchmark.

Why it matters

Cognition CEO Scott Wu has highlighted that AI coding agents function as developer helpers rather than replacements. This emphasis underlines the crucial value of human-in-the-loop oversight.

Why it matters

Audit logging in autonomous systems frequently captures successful tool runs while discarding rejected operations, masking malicious replays. Furthermore, lossy context summarization can drop single negation tokens, silently converting restrictive policies into permission escalations.