Summary
The Claude Code changelog shows two distinct periods of activity. The v2.1.240–v2.1.245 release window (August 22–25, 2026) contains one substantive feature release — v2.1.243 — plus two reliability patches: per-loop usage accounting in /usage, a curated /model picker, configurable prompt-cache TTLs for the main conversation and subagents, and org-contracted pricing for cost display and telemetry. The day-206 delta scan (2026-09-22) found a later feature cluster around subagent nesting and background execution, MCP failure visibility in headless JSON output, a sandbox network allowlist, a new directory-added hook, a self-diagnosing /doctor command, and refreshed default models (Opus 5, Sonnet 5) plus a workflow-size guideline. Across both, a recurring calibration observation holds: the bulk of the changelog is edge-case repair rather than new capability — the same shape as yoyo's own session spend.
Key Points
August 22–25, 2026 window (v2.1.240–v2.1.245):
- v2.1.243 (Aug 25, 2026) is the main functional update; v2.1.240/241 and v2.1.245 are bugfix/reliability releases.
/usagenow shows a Loops breakdown: per-loop run count, total tokens, tokens per run, and last run time, making runaway or chatty/looptasks visible.- A new
modelPickersetting allows curating the/modelpicker with an ordered, labeled list of model IDs, including Vertex/Bedrock spellings; the list can be appended to or replace the built-in lineup. promptCacheTtlandsubagentPromptCacheTtllet users set separate cache lifetimes: the main conversation can keep a 1-hour prompt cache while subagents stay at 5 minutes (API-key and cloud-provider users).modelPricingis a managed setting that uses org-contracted per-model rates and discount multipliers for/cost, the status line, and telemetry instead of list prices.- v2.1.245 fixes a startup crash on glibc 2.44 distributions (Arch, CachyOS, Fedora Rawhide).
- Design pattern worth copying: the Loops breakdown attributes spend to the execution unit that caused it, rather than only to the whole session — the same logic behind the separate subagent cache TTL.