safety plugins

by Aria Han

KERNEL is a Claude Code and Codex runtime for direct, decomposed execution with local memory and minimal hooks. Version 9.11.0 deploys the simplify deletion doctrine, preflight and verifier contract. Skills invoke as /kernel:<name> on Claude Code and $kernel:<name> on Codex; run /kernel:help first. Agent-readable summary: llms.txt.

by Dr.Kittimasak Naijit

Thai Token Optimizer v2.0 — adaptive Thai token optimization for Claude Code using lifecycle hooks, compact Thai policy, MTP-aware compression guidance, quality checks, and safety preservation.

by smrafiz

Shell-level guardrails for Claude Code. Command blocking, code security scanner, self-teaching, token economy, agent routing, MCP profiles, and session memory.

by widnyana

Denies git commands that stage protected paths, whole-tree adds, or rewrite history.

by widnyana

PreToolUse/Bash hook that blocks IaC write/destructive commands (ansible-playbook/ansible-pull without --check, ad-hoc ansible write modules, terragrunt/tofu/terraform apply-family including mise-wrapped forms), so agents never apply changes to real infrastructure.

by Karan Bansal

A one-file PreToolUse door-guard that bounces destructive Bash commands (rm -rf, DROP TABLE, curl|sh, force-push, device wipes) before they run. Blocks 45/45 known footguns, 0 false positives.

by ncaq

Rewrites `rm` (including trash-cli-compatible flags) to `trash` in Bash commands via PreToolUse hook.

by Jesse Vincent, forked by brunob54

Structured workflow skills for Codex — systematic debugging, TDD, brainstorming, code review, and 30+ expert workflows. Skills work on all platforms. Lifecycle hooks (routing, safety, reminders) require separate setup on macOS/Linux.

by SoDam AI Studio

초보 바이브코더를 위한 AI 진입점 — 계획 먼저·쉬운 검토·안전 기본 (Claude Code + Codex)

by Hermes Labs

A guided, project-local first experience for Claude Code that helps beginners start making something real.

by Clownware

A decision layer for Claude Code hooks. Classifies proposed tool calls against a policy you own and returns allow / ask / deny, using a fast System One model instead of a frontier LLM. Observe-only until you have looked at your own calibration table.

by Ruel Nopal

Two authority levels that separate deciding from doing. /planfirst investigates with read-only tools, probes the riskiest assumption, produces a plan in eighteen sections and STOPS. /proceed executes that plan, in sequence and within its scope, and reports against the acceptance criteria it wrote.

by Jordan Richlen

Before any call that transmits repo or user content off-machine (posting a comment, pushing a branch, calling an external API with file contents in the payload), state what is being sent and to whom — permission modes gate the call, this gates the content. Use whenever output leaves the machine to a destination the user didn't name in this task.

by dunnock

Conduct Fault Tree Analysis (FTA) to systematically identify and analyze causes of system failures using Boolean logic gates. Top-down deductive method for safety and reliability engineering. Use when analyzing system failures, evaluating safety-critical designs, calculating failure probabilities, identifying minimal cut sets, assessing redundancy effectiveness, or when user mentions "fault tree", "FTA", "system failure analysis", "minimal cut sets", "safety analysis", "failure probability", "AND/OR gates", or needs to trace failure pathways from top event to basic events. Supports qualitative structure analysis and quantitative probability calculations.

by Dongmin Yu

Seven 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 and shellcheck findings.

by Jordan Richlen

Rehearse the rollback before any irreversible action: name the specific restore path and demonstrate it works — never proceed on the strength of 'a backup exists'. Use before deletes, drops, force-pushes, migrations, or any action semver-gate classifies as MAJOR.

by Simon Heimlicher

Rust development workflow with testing, implementation, review, and architecture skills

by Simon Heimlicher

Rust development workflow with testing, implementation, review, and architecture skills

by Simon Heimlicher

Rust development workflow with testing, implementation, review, and architecture skills

by Javi

Review application security and trust and safety risks with evidence-backed findings and practical remediation guidance.

by tenfyzhong

Block destructive agent commands before they run.

by Hermes Labs

Blocks permanent-delete shell commands in Codex and redirects agents to a recoverable trash workflow.

by Hermes Labs

Recoverable deletion for coding agents: move files to trash, list them, and restore them with agent-trash.

by agentsmd

CODEX-CODING-SPEC — global Codex workflow and evidence guidance, with native hooks for selected safety/report checks and rule-specific opportunity telemetry.