Skip to content

Skills

Agent skills shipped by this repo, under skills/<name>/SKILL.md. They are the authoring-time half of the discipline: the checks enforce the conventions; the skills teach coding agents to write within them in the first place.

Every entry below is generated from skills/*/SKILL.md frontmatter — a new skill appears here the moment its SKILL.md is committed and jk-standards emit skills runs. Nothing on this page is hand-maintained.

doc-anti-drift

Write and maintain documentation under the jk-standards anti-drift discipline — doc lifecycle classes, drift maps, dated status claims, symbol-based references, test-cited behavioral claims. Use whenever creating or editing docs in a repo that has a jk-standards.yaml.

escape-hatch-discipline

Design and use lint/check suppressions correctly — every escape hatch in-band, greppable, and carrying a written reason. Use when suppressing any finding (linter, sanitizer, review scanner, doc check) or when adding a new check that needs an exemption mechanism.

sanitizer-ci-setup

Wire ASan/UBSan/TSan, fuzzing, and layered quality gates into CI for C/C++ projects — nightly sanitizer matrices, issue-dedupe notification, sanitizer-aware tests, strictness gradients. Use when setting up or auditing CI for a native-code project.

Vendor skills into a project with a skills-lock.json entry pointing at this repo, and install with the lock-file installer:

{
"version": 1,
"skills": {
"doc-anti-drift": {
"source": "JimAKennedy/jk-standards",
"sourceType": "github",
"skillPath": "skills/doc-anti-drift/SKILL.md",
"computedHash": "<sha256 of SKILL.md>"
}
}
}
python scripts/install_skills.py # → .agents/skills/
python scripts/install_skills.py --dest .claude/skills # Claude Code layout

The installer verifies each skill’s SHA-256 against the lock file, so a consuming repo pins skill content the same way it pins hook and workflow versions. Vendored skill directories are gitignored in consumers; project-authored skills stay tracked alongside them.