cc-agents-kit

Carry work across sessions and keep the context that carries it lean: hand off a session, export or log what it found, chart an effort too big for one sitting, and audit or garbage-collect your Claude Code config.

6 skills · hooks

What runs between "the code works" and "it is pushed": trunk setup, conventional-commit splitting, dependency vulnerability triage, spell-check and markdownlint triage, and babysitting CI until it is green.

6 skills

Nine defensive hooks: blocks recursive rm on $HOME, download-and-execute pipelines, secrets-file access, path-less formatter runs, credentials in the staged diff, and the two zsh quoting mistakes that fail silently; warns on stale lockfiles, shellcheck findings, and automatic security-review findings after a commit; masks credential-shaped values out of Bash output before the model sees them.

hooks

Temporary workaround for anthropics/claude-code#99704, to be removed once it is fixed: keeps user-facing text in the language set by `language` (Korean, Japanese, Chinese and other non-Latin scripts) by restating the rule after tool results, and asks for a restatement when a turn drifted into English.

hooks

A function-hook mod that reads each model request's own usage, shows the real context fill above the prompt, vetoes an auto-compact the advisor's double-counted usage would fire, and sends the handoff nudge once at 80%.

A function-hook mod that collects the files the agent reads, writes or names, and opens them in macOS Quick Look from /ql or a pane, in any terminal including Warp.