SKILL.md packages that extend Claude Code, Cursor, Copilot, and other AI agents.
Tags
claude-code-templates
Guides creating Hookify rule files (.claude/hookify.*.local.md) to detect patterns (dangerous commands, secrets, debug logs) and show warnings or block actions.
swarm-ai-safety
Research framework for simulating multi-agent systems to assess emergent risks, governance levers, and soft probabilistic metrics (toxicity, quality gap).
opendal-oli
Command playbooks and safe-operation patterns for the OpenDAL 'oli' CLI — listing, copying, moving, editing, and benchmarking data across local and remote profi
vcsdd-claude-code
Conventional commit and tagging conventions plus atomic staging rules for committing VCSDD pipeline artifacts to git.
skills
Classify chemicals and generate safe laboratory storage groupings and compatibility warnings (acids, bases, oxidizers, flammables, toxics) to help labs meet OSH
claude-skill-registry
Two-step Solana swap workflow: quote/simulate first, present clear summary, then execute only after explicit user confirmation to avoid accidental trades.
arifos
High-level autonomous governance workflow that outlines self-healing, self-optimizing, and constitutionally governed agent behavior for orchestration and govern
arifos
Legacy copy of an autonomous governance skill describing self-healing, self-optimizing and constitutionally-governed agent actions — archival version.
ISC-Bench
Benchmark template for evaluating spam-detection models using anchored spam campaign examples (contains harmful anchoring content).
spec-gen
Apply a confirmed refactoring plan safely: reads a refactor plan, applies changes as small, test-gated mini-developments, and stops on failures to avoid lost co
litellm-skills
Delete one or more organizations from a LiteLLM proxy with confirmation and safe checks.
agent-toolkit
Production-quality Bash scripting guide enforcing safe idioms (set -euo pipefail), ShellCheck compliance, and patterns for maintainable scripts.
ai-driven-development
Techniques and templates for crafting precise, reliable prompts and agentic instructions across LLMs.
mean-reviewer-skill
Generates an aggressive, exhaustive simulated peer review for stress-testing manuscripts and rebuttal workflows; includes mandatory license/acceptance gating.
agent-almanac
Practical, step-by-step guidance for cutting gemstones (cabochon or faceted): assessment, orientation, dopping, grit sequence, and basic faceting angles.
zorai
PreToolUse hook that blocks destructive git commands (push, reset --hard, clean, branch -D) so agents like Claude Code cannot run them without approval.
ISC-Bench
Benchmark and pipeline for evaluating model safety (Internal Safety Collapse) using the TVD framework; builds prompts, runs models, extracts harmful content, an
nonstop
Session-scoped autonomous work mode: runs a strict pre-flight risk assessment, dangerous-ops approval, and a blocker decision framework to let the agent continu
script-kit-next
Durable pattern and tooling for running bounded autonomous audit loops ("Go AFK for N hours") with cadence, budget, and safe stop conditions.
wizards-of-the-ghosts
Emergency containment skill: guidance and guarded procedure for immediate hard termination of processes, services, or accounts when graceful options are unavail
ros-skill
Control and monitor ROS/ROS2 robots over rosbridge (WebSocket) with CLI commands for topics, services, nodes, params, and actions.
awesome-omni-skills
Git-aware revert workflow for reverting tracks, phases, or tasks while preserving provenance and upstream workflow; includes safety checks and an execution plan
tandem-browser
Connect to a running Tandem instance (MCP or HTTP) to inspect, interact with, and automate actions in the user's real browser context while preserving safety an
