Hacker News
Claude Code 2.1.271 Adds `claude plugin eval`, a Scored Reproducible Test Suite for Your Own Plugins
The September 15 release adds `claude plugin eval`, which runs a plugin's eval suite against Claude Code and returns scored, reproducible results, plus `/output-style [name]` for switching presentation format mid-session and `CLAUDE_CODE_WORKFLOW_MAX_CONCURRENT_AGENTS` for raising the parallel agent cap. It also adds `OTEL_METRICS_INCLUDE_REPOSITORY` to tag telemetry with repository attributes, and fixes WebFetch hanging forever on servers that hold a response open without finishing it. The 2.1.270 hotfix two days earlier undid a regression where read-only git commands started asking for permission in long-running sessions.
↳ Follow the thread