Tanguychenier
- Updated May 21, 2026
- tanguychenier/better-call-claude
better-call-claude is a Claude Code marketplace plugin that installs five behavioral guidelines plus an optional shell guardrail hook so coding agents ask before assuming, stay in scope, and verify work for developers ti
About
Better Call Claude is a Claude Code marketplace plugin from tanguychenier that packages five agent-behavior principles—think before coding, keep changes small, stay in scope, drive toward testable goals, and verify against reality—plus synced CLAUDE.md and AGENTS.md drop-ins for cross-tool use. Install via `/plugin marketplace add tanguychenier/better-call-claude` or curl the rule files into your repo. The plugin skill loads the guidelines globally, and an optional PreToolUse hook blocks a narrow set of irreversible shell commands such as `rm -rf /` and fork bombs. Reach for it when agent diffs balloon, assumptions go unstated, or you want consistent agent discipline without writing custom CLAUDE.md rules from scratch.
- Five principles: think first, small diffs, in-scope edits, testable goals, verify reality
- Optional PreToolUse guardrail blocks rm -rf /, fork bombs, mkfs, and raw disk writes
- Ships synced CLAUDE.md, AGENTS.md, and GitHub Copilot instruction templates
- Install via Claude Code marketplace or curl rule files into any repo
- Cross-agent compatible: Claude Code, Codex, Copilot, Windsurf, Amp, Devin
Tanguychenier by the numbers
- Data as of Jul 7, 2026 (Skillselion catalog sync)
/plugin marketplace add tanguychenier/better-call-claudeAdd your badge
Show developers this marketplace is listed on Skillselion. Paste this into your README.
| Last updated | May 21, 2026 |
|---|---|
| Repository | tanguychenier/better-call-claude ↗ |
How do you stop AI agents from over-building code?
Install agent guidelines and a PreToolUse guardrail hook to reduce over-engineering, scope creep, and unverified 'done' claims in Claude Code sessions.
Who is it for?
Developers running Claude Code or cross-tool agents who want cleaner diffs, explicit assumptions, and verifiable success criteria on every task.
Skip if: Teams that already enforce behavior through a mature custom CLAUDE.md and do not need a generic five-principle baseline or shell guardrail.
What you get
Synchronized CLAUDE.md and AGENTS.md rule files, an installable skill, and an optional guardrail.sh hook wired through .claude/settings for PreToolUse blocking.
- CLAUDE.md
- AGENTS.md
- guardrail.sh hook
By the numbers
- Defines 5 behavioral principles across synchronized CLAUDE.md and AGENTS.md files
- Optional guardrail hook blocks a conservative set of catastrophic shell commands
Plugins in this marketplace
1 plugin - install individually after you add the marketplace.
Recommended Marketplaces
How it compares
Pick better-call-claude when you want a minimal, opinionated behavior baseline and optional shell seatbelt rather than a full multi-agent SDLC harness.
FAQ
What are the five Better Call Claude principles?
Better Call Claude defines five principles: Think Before Coding, Keep It Small, Stay In Scope, Work Toward a Testable Goal, and Trust Nothing Until You've Checked It. They ship in AGENTS.md and CLAUDE.md to reduce silent assumptions, bloated abstractions, and unverified completio
Does better-call-claude block all rm -rf commands?
No—better-call-claude's guardrail hook blocks only catastrophic targets like rm -rf on /, ~, $HOME, or top-level system directories, plus fork bombs and raw-disk writes. Commands such as rm -rf node_modules remain allowed to avoid false positives.
How do you install better-call-claude as a Claude Code plugin?
Run /plugin marketplace add Tanguychenier/better-call-claude then /plugin install better-call-claude@Tanguychenier inside Claude Code. The plugin loads the skill globally and installs the optional guardrail hook with the marketplace package.