Skip to content

v1.7.0

Choose a tag to compare

@nrzz nrzz released this 16 Oct 18:30
· 31 commits to main since this release
4d39153
  • A sixteenth and a seventeenth tool. claude-automode-doctor reads what Claude Code's auto mode and its classifier decided from your transcripts: who made each tool decision, the commands the classifier refused, and the refusals that were no verdict at all; its lint finds an autoMode list that drops the built-in rules because "$defaults" is missing. claude-mcp-check lists every MCP server Claude Code would load with the file it came from, starts each one the way Claude Code would and sizes its tools, and lints the configuration. Both are on-demand tools on the setup page (auto mode doctor with report, show and lint; MCP check with show and lint; none of the buttons writes a file or starts a server) and plugins in this marketplace with one user-only skill each.
  • Always-on tokens: the fourteen plugins come to about 212 by Claude Code's own estimate, over the old limit of 200, only because the estimate counts the descriptions of user-only skills, which the model is never sent (each new tool's README shows its request is the same size with and without the plugin). The gate is the check that already ran beside it: no request the fake API receives may mention a user-only skill. The estimate stays as a tripwire, under 250.
  • Website and README: seventeen tools, fourteen plugins, the two cards and rows, over 6,900 tests, and the preview images say seventeen.