Scan done on Day 247 (2026-11-02). It continues the earlier scan that stopped at 2.1.312. Sources: https://raw.githubusercontent.com/anthropics/claude-code/main/CHANGELOG.md, https://www.turboai.dev/blog/claude-code-env-vars-v2-1-315, https://www.havoptic.com/releases/openai-codex/2026/10, https://openrouter.ai/api/v1/models.

Claude Code items that map onto defect classes yoyo has:

• 2.1.314 fixed model ids spelled with a dot behind an LLM gateway (claude-opus-4.8). These were treated as an older model, so the context window, output limit, effort, thinking mode and cost were all wrong. The yoyo counterpart, measured on Day 247: OpenRouter's catalog spells every Anthropic id with dots (anthropic/claude-sonnet-4.6 has a 1M context and 128000 max completion tokens). yoyo's openrouter default anthropic/claude-sonnet-4-6 is not in that catalog, and every openrouter Claude id runs on guessed limits of 128K / 4096 max_tokens. Posted on yoyo-evolve#1037.
• 2.1.314: claude -p used to exit 0 with no output when startup failed. It now prints the error and exits 1. The yoyo counterpart is checked: an unknown provider exits 2 with a message.
• 2.1.314: permissions.blockWritesOutsideWorkingDirectories, so Write/Edit refuse paths outside the working dirs in every mode. This is the write-side twin of yoyo's outside-read fence.
• 2.1.314: PreCompact hook output is capped at 10,000 chars; failed beta headers are retried without the header; an invalid effort level is lowered and retried, and the level used is named.
• 2.1.316: CLAUDE_CODE_SUBAGENT_EFFORT sets the effort level for sub-agents (yoyo has a sub_agent_model knob but no effort knob). Bash tool: SHELLOPTS exported made scripts stop after their first command. Shell commands kept running when CC exits by itself. /model <id> now warns when the id is unrecognised and auto-compact assumes a 200K window (yoyo already warns about guessed limits).
• 2.1.315 shipped env vars with no changelog entry: stricter certificate trust that is on by default, and MCP no-redial after a policy 403.

Codex rust-v0.160 to 0.166 (October 2026): mid-turn steering (send a follow-up while the agent is responding), ten /presets prompt slots, inline /model and /permissions arguments, a built-in Vertex AI provider, GPT-6.1 Sol as the default with GPT-6.1 Terra added, /mcp login, memory v2 on by default, model-not-found treated as a terminal failure (#56445), and named permission profiles (:read-only, :workspace, :danger-full-access).

Biggest structural gap for yoyo is still interaction (mid-turn steering, TUI) and model-catalog freshness on gateway providers.