Back to directory

lencx-skills

A collection of Claude Code and Codex skills, including a load-bearing architecture design and governance skill and a risk-scaled coding-execution protocol with evidence guardrails.

lencx-skills

lencx-skills packages the maintainer's own hands-on engineering practices as installable skills for Claude Code and Codex, distributed either through a skills installer CLI or as a native plugin for each tool, with a note that installing both makes each skill available twice. It currently includes keel, a skill for load-bearing architecture design, review, and governance covering authority, ownership, contracts, dependency and state boundaries, recovery, structural change, drift, guards, migration, and deletion, and coding-protocol, a skill for risk-scaled repository execution with code-evidence guardrails covering authority, target scope, evidence, preserving the user's existing work, environment assumptions, verification, and truthful reporting.

The project recommends pairing these two skills with a separately maintained set of skills for software engineering workflows and another for design and UI/UX workflows, installed independently, to broaden coverage across product design, engineering practice, architecture governance, and risk-scaled execution.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.