Kimi Code Integration (Native Claude Skills Spec)

Uses the native Claude skills specification via extra_skill_dirs in ~/.kimi-code/config.toml.

Prerequisites

• Jira auth configured — per-request Basic auth from .jst/jira-sdlc-tools.local.env, no login step (see jira-api-reference.md §9)
• gh (GitHub CLI) authenticated
• .jst/jira-sdlc-tools.env and .jst/jira-sdlc-tools.local.env in your project root — see project-config.md
• Kimi Code CLI installed

Install / Wire-up Steps

Kimi Code reads Claude-spec skills natively from paths declared in extra_skill_dirs at the top level of ~/.kimi-code/config.toml. No copy step, no per-skill adaptation file.

1. Add the plugin's skills directory to ~/.kimi-code/config.toml:

   extra_skill_dirs = [
       "</PATH>/plugins/jira-sdlc/skills"
   ]

2. Replace </PATH> with the absolute path where this plugin lives on your machine:

   • Installed via marketplace: ~/.claude/plugins/cache/jira-sdlc-tools/jira-sdlc/<version>/skills — pin to a specific versioned directory (Kimi Code does not auto-update this path; re-point it after a plugin upgrade)
   • Local clone: the absolute path to plugins/jira-sdlc/skills

   _shared sits alongside the three skill folders (no SKILL.md, so Kimi Code does not load it as a skill), and the relative references inside each SKILL.md stay intact because they point within the same tree.

3. Start a new Kimi Code session — skills are discovered on startup. The three skill descriptions appear in the /skill listing.

Invoking the Three Skills

Kimi Code registers skills as slash commands in the format /skill:<name> — there is no plugin namespace:

• /skill:jira-task-assigner — break down a task into Jira issues with branches + worktrees
• /skill:jira-task-executor — implement an issue end-to-end from its worktree
• /skill:jira-task-reviewer — review sub-task PRs from the parent worktree

This is a bare-name invocation, not the /jira-sdlc:… namespace Claude Code uses. The cross-references inside the skill bodies (/jira-sdlc:jira-task-executor, etc.) are prose pointers — on Kimi Code the user reruns the bare /skill:jira-task-… command instead. Verified in a live Kimi Code session — this doc was written from a Kimi Code run that loaded the skills via extra_skill_dirs and invoked /skill:jira-task-executor.

Platform-Specific Caveats

• disable-model-invocation: true — honoured natively (Verified). Kimi Code recognizes this frontmatter field as an alias of disableModelInvocation: when true, the skill is hidden from automatic model invocation and must be called explicitly via /skill:<name>. All three jira-sdlc skills ship with this flag set deliberately (explicit invocation only), and Kimi Code honours it without any per-skill override or adaptation file. Verified — this very run exercised /skill:jira-task-executor with disable-model-invocation: true in its frontmatter, and the skill activated only on the explicit slash command.
• Skill precedence — Extra tier. extra_skill_dirs lands skills at the Extra tier: Project > User > Extra > Built-in. A project-local .kimi-code/skills/ or user-level ~/.kimi-code/skills/ skill with the same name overrides one loaded from extra_skill_dirs. This matters when the plugin ships a future skill that a project also maintains independently. Verified from Kimi Code docs.
• KIMI_CODE_HOME does not move extra_skill_dirs-sourced skills (Verified). If you set $KIMI_CODE_HOME, the user-level skills scan directory (~/.kimi-code/skills/) moves with it, but extra_skill_dirs entries point at absolute paths outside that home — they are unaffected. Only the user-tier skills directory is relocated. Verified from Kimi Code docs.
• Plugin-cache path pins to a version. Marketplace-installed plugins live under ~/.claude/plugins/cache/jira-sdlc-tools/jira-sdlc/<version>/. Kimi Code reads the skills directly from that path and does not auto-update the extra_skill_dirs entry. After a plugin upgrade (new version directory), re-point extra_skill_dirs to the new version path and restart. The manual-copy drift caveat of Mechanism B does not apply — the extra_skill_dirs path always reads the installed files as-is. Verified (observed with ls ~/.claude/plugins/cache/jira-sdlc-tools/jira-sdlc/).
• Frontmatter field name aliasing. Kimi Code accepts disable-model-invocation, disableModelInvocation, and disable_model_invocation as equivalent. Jira-sdlc ships with the hyphenated form disable-model-invocation (Claude Code canonical), and Kimi Code's docs list it as an explicit alias. Verified from Kimi Code docs.
• Skill nesting limit — 3 levels. Kimi Code allows up to 3 nested skill invocations; beyond that, invocations are terminated. The jira-sdlc skills do not nest beyond 2 levels (executor calls ensure_local_env.sh etc., but those are scripts, not skills). Verified from Kimi Code docs — no impact.