Collapsing every copy of one kind of thing (tables, UIs, utilities, hooks, APIs, renderers) into a single canonical one. Use when asked to unify, merge, converge, or 'make one X used everywhere', or before migrating callers onto a new canonical piece. NOT for docs (use docs).
インストールする前に、エージェントに与えられる指示の中身を確認できます。
The rich-content census took three rounds and each one still found surfaces the last missed. Every step below ends on a check; skipping one is the failure.
Create one list file the first minute (<project>/LIST.md or the project's register). Every instance
you find, in any round, by anyone, goes on it immediately: what · where (file / table / route / repo)
· how found · status. Never rebuild it from memory or from a summary. A find that is not on the list
was not found.
Done when: the list exists and every later step writes to it.
Keyword grep is round one, never the last round. Each round uses a DIFFERENT method:
Where copies do the same job differently, list every way on the list and pick a winner per capability. Build the canonical one CLEAN from the winners: never promote the best existing copy and patch it.
Done when: a round using a new method adds nothing to the list.
Work the list top to bottom. Each item: move it, verify it on its real surface, mark it done with the proof. No bulk "all migrated" claims. Before moving each item, screenshot it in place and apply canonicalize without destroying: an ordinary copy is replaced; a copy that is visibly better in its spot is adopted into the canonical piece (feature, option or new primitive) after one simple question to Arman with before/after links. Only then is the old copy deleted, not left beside the new one.
Done when: every row is done-with-proof or explicitly out of scope with a reason.
When you think you are done, dispatch agents that did NOT build it and do NOT get your list: they search every codebase from scratch for anything still doing the job the old way. Every find goes on the list and back through step 3.
Done when: an independent sweep returns nothing new.
Charge at least one agent to think outside the box: what does this same job under another name? (The assistant answer is a note. A flashcard front is rich content. A system prompt is a document.) Anything that fits joins the list.
Done when: that agent's finds are on the list and migrated or ruled out with a reason.
Add a guard that fails on any NEW use of an old copy, with a baseline that only shrinks, proven
failing-then-passing (forcing-function-tests).
Done when: the guard is in the release gates and its self-test goes red then green.
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Admin debug system wiring for the floating AdminIndicator's live debug data and Copy Full Context. Use when adding debug visibility to a route or feature, wiring useDebugContext or the debug panel, capturing console errors, or enabling the copy-context workflow for a page.
日本語の概要は準備中です。原文の説明を表示しています。
Compact two-icon Copy / Copy-for-AI controls (components/agent-copy). Use when adding copy buttons to a row, card, list, or record; merging duplicate Copy/JSON/AI controls; continuing the copy rollout; or writing a Copy-for-AI payload. NOT for markdown content actions (use rich-document-actions).
日本語の概要は準備中です。原文の説明を表示しています。
Disclosing a surface's existing fixed AI jobs in the shell's top Agents menu. Use when a page, panel, overlay, or window already runs a mandate behind a button, assist, automatic action, or mode; when the agent-disclosure guard names a file; or during a surface check.
日本語の概要は準備中です。原文の説明を表示しています。
The watch-fix-rerun method for making an existing platform agent or automated process efficient and correct: baseline its ledger, run one unit yourself, fix the class behind every wasted call, rerun, record. Use when asked to improve, optimize, watch, or 'make efficient' an agent, a sandbox session, a sync job, or any recurring automated process.
日本語の概要は準備中です。原文の説明を表示しています。
Redux state for agent execution and firing agent shortcuts. Use when editing features/agents/redux/, building agent UI, creating a conversation, touching assembleRequest or NDJSON stream state, wiring agent-state selectors, adding a per-conversation capability, or triggering a shortcut from a button, menu, or code (useShortcutTrigger, launchAgentExecution).
日本語の概要は準備中です。原文の説明を表示しています。
The provision for a call site: the exhaustive menu of values that place in the code can realistically produce. Use before creating or fixing a mandate or agent that reasons about more than its own input, when deciding whether an agent can answer at all with what it is sent, or when its output 'looks right' but nobody checked it could be. NOT for building the agent (use create-agent).
日本語の概要は準備中です。原文の説明を表示しています。