Skill Claude CodeCodex
Run or resume NMG's OmniMemEval user-memory benchmarks through the config-driven shared workflow. Use for LongMemEval, LoCoMo, BEAM, PersonaMem v2, HaluMem, benchmark regression checks, or result continuation.
Skill Claude CodeCodex
Maintain this repository's design, decision, experiment, post-mortem, bilingual, TODO, and operating documentation. Use whenever a change adds or revises behavior, architecture, process, evaluation evidence, public instructions, an escaped failure and its guardrails, or when documentation is reorganized, translated…
Skill Codex
Use NMG as durable memory or temporary multi-Agent coordination when a task may depend on prior user facts, preferences, constraints, decisions, project state, events, reusable experience, or a shared task blackboard; when the user asks to remember or recall something; or when an Agent must start, query, and safely…
Skill Claude CodeCodex
Modify, test, and commit this repository safely. Use for any NMG code, test, CI, packaging, or repository-tooling change.
Skill Claude CodeCodex
Reduce cyclomatic complexity in NMG code (CodeFactor Complex Method, or local tools/complexity-gate.ts). Use when a method is flagged as too complex, or before committing code that grows a method's branching.
Skill Claude CodeCodex needs its repo
Use when a task needs a throwaway script, ad-hoc analysis, or probe in this repository — how to write one that is right (flags, no silent zeros, a self-check, provenance in the output), and where the ready-made parts are if you would rather import them.
Skill Claude CodeCodex
Make a declared design assertion checkable and traceable. Use when adding or changing a contract assertion, choosing how to verify a behaviour, or when a behaviour has no expected output.
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: