← ClaudeAtlas

ki-engineeringlisted

Use to audit or conform the shared Knowledge Islands TypeScript/Bun engineering standard: comprehension-first modularity and reuse; architectural-boundary testing; package scripts, tsconfig, Biome, and toolchain consistency. Triggers: "audit our engineering standards", "is this code too DRY", "are tests at the API boundary". For repository configuration use `ki-repo`; Markdown/TOML style use `ki-authoring`; MCP specifics use `ki-repo-mcp`.
knowledgeislands/ki-agentic-harness · ★ 0 · AI & Automation · score 76
Install: claude install-skill knowledgeislands/ki-agentic-harness
# Knowledge Islands engineering standard You are applying the **Knowledge Islands engineering standard** — the shared code-design and software-engineering toolchain every TypeScript/Bun repo in this work builds on. It is the build/test twin of `ki-authoring`: that skill owns _how we write_ (Markdown/TOML style); this one owns _how we structure, build, lint, and test_. `ki-skills` owns the governance-skill enforcement framework that this skill follows. This is a **standard, base-agnostic governance skill**. It hard-codes no single repo; it applies to any repo carrying a `[skills.ki-engineering]` table in its `.ki.toml`. The active repository set is local supporting evidence, not a normative count: derive it from the current configured collection during REFRESH. How this skill sits alongside the other skills, and where it must not overlap them, is documented once in the ki-agentic-harness `README.md`. ## What this skill owns 1. **Code design and the common toolchain** — comprehension-first modularity and reuse, plus the baseline every TS/Bun repo meets and capability conditionals that fire only when a repo opts into a capability. The full, quotable standard is [the engineering standard](references/standards-engineering.md); the line-by-line items are in [the rubric](references/rubric.md). 2. **The governance-skill rubric model** — the shared mechanism for turning any standard into structured criteria, focused evidence, phased audit and conform execution, and canonical check