Browse reusable Agent Skills, each with a clear purpose and practical guidance.
Audit World of ClaudeCraft Codex support for current conventions, model neutrality, instruction quality, agent safety, skill design, hooks, CI integration, and documentation drift. Use when reviewing or modernizing AGENTS.md, .codex, .agents, Codex workflows, Codex-facing scripts, or the repository AI architecture.
Use when planning, debugging, tuning, evaluating, exporting, or deploying public Nemotron `embed`/`rerank` retrieval recipes.
Troubleshooter for agentic-RL training, evaluation, and experiment design on LLM agents (single or multi-agent, multi-turn, tool-augmented). Routes a user's symptom to fixes anchored in the corpus. TRIGGER when: user is training, evaluating, or designing experiments for an RL-trained LLM agent; symptoms like reward not moving, eval flat, KL/entropy/length blow-ups, retokenization drift, tool-call parse failures, credit assignment, async-rollout staleness, judge inconsistency, benchmark contamination, pass@k vs pass@1; choices about ablation, baseline, framework, algorithm, reward, or data curation; user names GRPO, PPO, DAPO, veRL, OpenRLHF, slime, AReaL, RAGEN, or similar. SKIP: generic supervised LLM fine-tuning with no RL component; classical RL theory or tabular RL; non-LLM agents. Distilled from the AgentsMeetRL awesome list, snapshot 2026-06-20.
为了解决子智能体（Sub-agent）的上下文问题，逐步优化上下文获取的模式
建议在逻辑间隔处进行手动压缩（Manual Compaction），以在整个任务阶段保持上下文（Context），而非随意的自动压缩。
Audit your OpenClaw setup for token waste, context bloat, and cost optimization opportunities
Inspect and manage the cross-session engineering memory that Memanto maintains for your Claude Code skills. Use when the user asks what Memanto remembers, wants to see their engineering profile, manually recall context for a skill, or store a decision. The automatic lifecycle hooks handle capture/injection on their own — this skill is the manual control surface.
Create or refresh project-local DSL skills under `.agents/skills/<dsl-name>-dsl/` for one or more MPS languages after discovering concepts, sandbox examples, JSON blueprints, references, or gotchas.
Operational constraints for Scion agents running in containerized sandboxes. Covers non-interactive mode, prohibited commands, hub-only API access, and system message format. Complements the scion CLI reference and messaging skills.
Adding a new MoonBit agent to a Golem component. Use when the user asks to create, add, or define a new agent type in a MoonBit Golem project.