Kohei-Wada

Use when user wants to refresh one or more profiles end-to-end - runs nemawashi-collect followed by nemawashi-analyze for each named target (or --all), dispatching one profile-updater agent per target with throttled parallelism.

Use when starting any conversation - establishes how to find and use interpersonal communication skills

Use when user wants to analyze a person's behavioral patterns - reads profile data and adds psychological analysis with actionable DO/DON'T communication rules

Use when user wants to upgrade legacy profile data to a newer on-disk format - lists available migrations, presents the candidate profile counts, and applies each migration via LLM-driven transformation

Use when user needs help replying to someone, crafting a message, or deciding what to say - references profiles to tailor communication strategy

Use when user wants to view, show, list, or display profile data for one or more people - read-only display of profiles stored in PROFILE_DIR

Use when user wants to check which profiles need re-analysis - scans all profiles for staleness and new unanalyzed data

Use when the user wants to file a GitHub issue against this repo from free-form feedback or an idea surfaced in conversation — drafts a house-style body, surfaces duplicates, and files via gh after explicit confirmation.

Use when the user wants to wrap up the current Claude Code session by writing what was worked on to today's daily note in the vault, so the next session can pick up context. Drives recap.manual_recap to upsert the per-session two-layer recap block (Timeline + KPT).

Use when the user wants to turn vault knowledge into a published blog post — gathers the relevant permanent notes, shapes the post in a dialogue (no stored draft), masks PII, and emits it into the blog repo per that repo's own conventions. Triggers on "blog 化して", "記事化して", "publish this as a post", "ブログにして", "harvest these notes into a post".

Use when starting any conversation - establishes how to find and use long-term knowledge curation skills that read, write, link, and prune the user's external knowledge base

Use when an existing MOC and an existing child note need to be linked (bullet under the right MOC sub-heading, optionally a reciprocal back-link from the child). Atomic graph-edge insertion only — no surrounding prose, no section creation. Pairs with garden-water (content edits) and garden-plant (new notes).

Use when a reusable insight, decision, lesson, or piece of durable knowledge surfaces in conversation and should be captured to the user's long-term vault - proposes a new note formatted per the vault's own README conventions

Use when one or more named existing vault notes need to be removed — archived by default (git mv into the vault's documented archive folder, preserves history), hard-deleted only on explicit request. Surfaces inbound-link warnings but never auto-rewrites links. Pairs with garden-survey (find candidates) and garden-water (clean up links afterwards).

Use when an existing vault note needs new information, an added link, refined wording, or a metadata fix (tag / frontmatter field). Operates as a minimal diff on the existing note, never a wholesale rewrite. Pairs with garden-plant (new note) and garden-prune (removal).

Use when the user or another skill needs to search, list, or query the vault — by text, tag, frontmatter field, or folder. Returns concise structured results (path, tags, snippet) without modifying anything.