boundary-bbcr-fallback
Execute automatic BBCR (Collapse-Rebirth Correction) when knowledge boundaries are exceeded or reasoning fails.
Set up a file watcher that triggers Claude on file changes
$ npx -y skills add qdhenry/Claude-Command-Suite --agent claude-codeHow it fires
How this command gets triggered: by you, by Claude, or both.
/watchContext preview
What this command does when you run it.
Set up a file watcher that triggers Claude on file changes
description: Set up a file watcher that triggers Claude on file changes argument: Watch path (file or directory)
Set up a chokidar-based file watcher for: `$ARGUMENTS`
1. Ask the user the following questions interactively (use AskUserQuestion):
2. Verify the watch path `$ARGUMENTS` exists. If it doesn't, warn the user and ask if they want to proceed anyway.
3. Ensure `scripts/watch-files.mjs` exists. If not, inform the user it needs to be created (it should already exist as part of the file-watcher skill).
4. Ensure `chokidar` is in devDependencies. If not, run `bun add -d chokidar`.
5. Ensure `package.json` has the `watch:files` script. If not, add it:
"watch:files": "node scripts/watch-files.mjs"
6. Run a smoke test:
7. Print the final command the user can run:
node scripts/watch-files.mjs <path> '<prompt>' [--glob '<pattern>'] [--debounce <ms>]
See `.claude/skills/file-watcher/SKILL.md` for full documentation, examples, and troubleshooting.
A comprehensive development toolkit designed following Anthropic's Claude Code Best Practices for AI-assisted software development.
Repo: qdhenry/Claude-Command-Suite
Execute automatic BBCR (Collapse-Rebirth Correction) when knowledge boundaries are exceeded or reasoning fails.
Analyze semantic position relative to knowledge boundaries to prevent hallucination and identify uncertainty zones.
Generate a visual heatmap of knowledge boundaries showing safe zones, risk areas, and semantic coverage.
Evaluate the current risk level and provide detailed analysis of potential hallucination or reasoning failure.
Find and construct semantic bridges to safely navigate from current position to target concept without crossing dangerous boundaries.
Takes an input prompt and returns ONLY a token-optimized version that preserves meaning while minimizing token count. Based on LLM tokenization principles:…