claude-code

Lessons that fire at the moment of action: reminders, questions, or blocks attached to matching tool calls.

A Claude Code plugin that fires a toast when a turn rebuilt the prompt cache, showing tokens written, turn cost, and the likely cause (idle or model change).

A Claude Code mod (Claude Code 2.1.287+) that draws a band above the prompt with the messages waiting for this session on the Agent Team OS bus.

A Claude Code function-hook mod that flags hardcoded user home paths (/home/<user>, /Users/<user>) written into files and nudges $HOME / ~ instead. Machine-local (*.local.*) files are exempt, and the note is non-blocking.

Removes MCP server descriptions that are not useful in the current directory according to directory rules, saving context; rules are configured in userConfig under rules, and it does nothing when no rules are configured.

See what your skills and MCP servers put in front of Claude, which tools it called, what it read back, and what it did next. Observe-only; writes local logs only.

A Claude Code plugin marketplace shipping branch-usage, which aggregates token usage and cost per git branch across sessions, models and subagents, with a live pane, a cost line above the prompt, a /branch-usage report, a get_branch_usage tool, and an optional automatic cost comment on the branch's pull request.

A searchable Claude Code model picker (/ms) for setups with very long model lists, leaving the built-in /model command unchanged.

Shows which files Read opened, which lines, and what each search looked for, on the folded tool line

A Claude Code mod that shows one line per running subagent above the prompt, including model, task, minutes and latest tool call.

A Claude Code mod that shows how much time is left on the prompt cache, so you can keep a session going (or compact it) before the cache expires.

A Claude Code plugin built with function hooks that shows each turn's typed prompt alongside Claude Code's last answer for that turn in a side pane via the /conversation command.