Skill UI

The Agent Harness provides a robust, verifiable loop structure for driving complex goals across specialized domains. It converts a high-level objective into a verifiable task plan, executes tasks sequentially using domain-specific tools, and rigorously verifies every step with machine-run checks. The system includes mechanisms for retries, budget management, and mandatory human escalation, ensuring no goal is closed until all progress is verifiably complete or explicitly waived.

A comprehensive utility for managing complex cmux topologies, including workspaces, panes, and surfaces. This tool is essential for inspecting, controlling, and monitoring multiple AI coding agents running in parallel within a macOS terminal environment. It supports advanced functionalities such as browser automation and real-time status notifications, optimizing developer workflows.

Execute discrete, bounded coding, review, or verification tasks using the Codex CLI as an isolated subagent. This tool is ideal for tasks requiring explicit file ownership, such as isolated refactoring, feature implementation, or independent code verification, without relying on continuous conversational context. It runs in a safe sandbox environment.

This guide details the comprehensive workflow for distributing a single AI skill across various local agent skill directories (Codex, Claude, Pi, Hermes). It addresses canonical locations, symlink management, and specific synchronization commands to ensure that all agents detect the latest skill version while correctly navigating common development pitfalls.

This skill is designed to mitigate false-positive safety triggers from strict AI classifiers (e.g., cyber, bio/chem, reasoning). It operates by making minimal, surgical edits to sensitive phrases, abstracting away specific domains or actions while preserving the original benign intent and structure. It ensures the prompt remains policy-compliant without losing its core purpose, making it ideal for academic or defensive inquiries.

This skill helps developers create highly focused, folder-specific documentation (`CLAUDE.md` and `AGENTS.md`). It systematically analyzes all files in a target directory to extract project purpose, operational constraints, naming conventions, and critical decisions. This deep context ensures that future AI agents (like Claude or Codex) operate with localized knowledge, preventing them from relying on global, generalized instructions and improving their accuracy in that specific domain.

Provides deep semantic code intelligence for the Go language, utilizing the official `gopls` language server. Key features include precise go-to-definition, finding references, analyzing call graphs, performing safe refactoring (rename, extract, inline), generating diagnostics, and formatting code. It is essential for navigating, understanding dependencies, and maintaining large Go codebases.

This skill guides the safe, systematic, and at-scale refactoring of existing Go codebases. It ensures that structural changes maintain observable behavior by following a rigorous process: mapping blast radius, strengthening safety nets (tests), applying small tool-driven transforms, and committing changes atomically via staged PRs. Ideal for cleaning up large functions, resolving code smells, renaming packages, or tackling complex architectural improvements in Go.

This skill automates the commit and push process for local skill changes. It should be used after a skill has been fully validated and updated, ensuring the canonical copy of the skill is published to the configured remote GitHub repository. This is essential for sharing or finalizing skill updates.

WGM is a robust, agentic methodology designed to transform vague or rough requirements into working, validated software. It enforces a disciplined workflow across six stages: deep alignment (Grill), structured planning, deterministic iterative building (Loop), mandatory holdout-scenario judging, and final documentation audits. It ensures that the resulting code converges against rigorously tested acceptance criteria rather than relying on single-pass generation.

This skill builds and maintains a persistent index and dependency graph for all Architecture Decision Records (ADRs). It handles both v3-style and YAML plugin-style ADR formats, persisting records into the adr-patterns namespace and relationships (like supersedes, amends) into adr-edges. It is essential for bootstrapping ADR tracking, synchronizing the knowledge graph, or ingesting records from external projects, ensuring a single source of truth for architectural history.

Generates a precise, evidence-based installation plan for a project's component catalog (ECC). Instead of loading a full, noisy bundle, this skill analyzes the repository's actual code structure, dependencies, and CI/CD evidence to classify skills, rules, and commands into 'DAILY' (essential) and 'LIBRARY' (referenceable) buckets. It ensures only necessary components are active, optimizing the development environment.