Skill Claude CodeCodex
Assembles high-signal repository context for non-trivial implementation, debugging, refactoring, code review, and repository investigation requests. Activates on requests needing fundamental context before Claude Code acts.
Skill Claude CodeCodex
This skill helps Claude perform evidence-based debugging and empirical diagnosis to identify root causes of coding issues systematically.
Skill Claude CodeCodex
This skill helps Claude evaluate AI prompts by defining evaluation contracts, building test matrices, and analyzing outputs for quality assurance.
Skill Claude CodeCodex
Enforces a structured protocol for detecting and responding to prompt injection attempts in LLM applications. Use when building an input-sensitive app, auditing for security vulnerabilities, or operating a model that interacts with user-generated content.
Skill Claude CodeCodex
This skill helps Claude optimize user prompts for clarity, technical accuracy, and effectiveness before sending them to an AI system.
Skill Claude CodeCodex
Architects the system prompt, context budget, and tool contract of an AI agent as one system. Use when designing a new agent prompt, reviewing an existing agent configuration, or debugging an agent that loops, mis-selects tools, or stops without a clear result.
Skill Claude CodeCodex
Triages how much change a prompt actually needs before rewriting it, then applies only the moves for that tier. Use when a user asks to improve, optimize, clean up, or rewrite a prompt and you must avoid both over-editing a good prompt and under-editing a broken one.
Skill Claude CodeCodex
Decides whether delegating work to a subagent is worth its cost, picks the right agent shape (fork vs. fresh vs. no delegation), and scopes the dispatch so it doesn't spend more context than it saves. Use when an AI coding agent is about to spawn a subagent, fork itself, or launch a parallel wave of agents, and needs…
Skill Claude CodeCodex
Audits whether the working context of a prompt or agent run has drifted across turns, retrieved sources, edits, and decisions, then reconciles the context before the next model action. Use for long-running prompt workflows, iterative agent runs, prompt version changes, or any task where stale, contradictory, or…
Skill Claude CodeCodex
Audits an existing evaluation pipeline (LLM judges, assertions, scorers, repair or regression gates) for silent failure by probing it, not by reading the code. Use when a prompt eval suite, LLM-as-judge setup, or score-gated workflow already exists and the question is whether its numbers can be trusted, especially…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: