docs: teach the two axes, not the one word

SKILL.md is the file an agent reads before it writes anything, and its
"Which origin you are talking to" section taught the coarse question. It
now teaches both: which tracker (jira | linear | gadak) and where it runs
(local | remote), with the paired workspace named as the case worth
remembering — gadak, remote, no Atlassian credential anywhere. `kind` is
still documented, as the older field to fall back on when a gadak is too
old to send the pair.

That section also said the durable file is `origin/issuetap.yaml`. It has
been `.db` since the persist migration; the sentence was simply stale.

README (both), CONCEPT, MIRROR, INSTALL, CONFIGURATION, NETWORK, FAQ,
RECIPES, AGENT_SETUP, SECURITY, the specs, the site and this repo's own
CLAUDE.md follow. `docs/decisions/` and CHANGELOG do not: those are
history, and history is not revised.

The `standalone-jira:` / `standalone-confluence:` id prefixes in
data-model.md stay exactly as they are — they are stored bytes, not
vocabulary.

GDK-1282 (epic GDK-1278). Gates: go, doc-checks.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Author: midagedev

CLAUDE.md

@@ -8,19 +8,19 @@ hard-won 목록)와 `AGENTS.md`(스키마·쿼리)가 원본이다.
 
 - **미러는 버려도 되는 캐시.** 원본은 항상 Jira — 그 Jira가 Atlassian
   Cloud든, gadak이 함께 들고 다니는 아주 미니멀한 셀프호스트 Jira
-  (`issuetap`, standalone 워크스페이스)든. 어느 쪽이든 미러는 origin에서 다시
+  (`issuetap`, gadak origin)든. 어느 쪽이든 미러는 origin에서 다시
   만들 수 있고, **gadak 자신은 원본을 보관하지 않는다.** gadak에만 존재하는
   원본 데이터를 만드는 변경은 금지 — 예외는 `local.db`(방문·검색 기록)와
   저장된 뷰이고, 그것들은 export 가능해야 한다.
-- **영속은 origin의 몫이다.** standalone 워크스페이스의 영속 공간은
+- **영속은 origin의 몫이다.** gadak origin 의 영속 공간은
   issuetap의 persist 파일이며 미러가 아니다. 그래서 백업 대상은 `gadak.db`가
   아니라 그 파일이고, **워크스페이스는 origin 하나에 묶인다** — origin을
   바꾸는 것은 설정 편집이 아니라 새 워크스페이스다. 이 조항을 어기는 것이
   "다른 트래커를 조용히 가리키게 하는" 부류의 결함이다.
 - **쓰기는 전부 origin(Jira)을 통과**한 뒤 미러 갱신. 미러에 직접 쓰는 API를
   열지 않는다. 위키도 같은 규칙이다 — 페이지 쓰기(생성·편집·코멘트,
   GDK-380/381/382)는 origin.Wiki를 통과한다: connected는 Confluence REST,
-  standalone은 issuetap의 Confluence API(미러 직접 쓰기가 아니다).
+  gadak origin 은 issuetap 의 Confluence API(미러 직접 쓰기가 아니다).
 - **아웃바운드 없음.** 텔레메트리 금지. 나가는 요청은 사용자가 설정한
   origin(Atlassian 사이트·Linear: api.linear.app/uploads.linear.app), GitHub
   릴리스 버전 체크(설정으로 끔), 페어링한 home serve, loopback뿐

README.ko.md

@@ -37,7 +37,8 @@ gadak은 Jira *그리고* Confluence를 로컬 SQLite 파일 하나로 미러링
 
 연결된 사이트에는 [API 토큰](https://id.atlassian.com/manage-profile/security/api-tokens)
 하나가 필요합니다. 토큰 하나가 같은 사이트의 Jira와 Confluence를 모두
-커버합니다. Standalone 워크스페이스는 Atlassian 계정 자체가 필요 없습니다.
+커버합니다. origin 이 gadak 자신의 트래커인 워크스페이스는 Atlassian 계정이
+아예 필요 없습니다.
 
 **무엇을 미러링할지는 직접 고릅니다.** 위키는 요청하기 전까지 꺼져 있고,
 켤 때는 스페이스를 지정합니다(`gadak init --spaces ENG,PROD`, 또는
@@ -79,7 +80,8 @@ gadak serve
 ```
 
 트래커를 떠날 때는 데이터를 들고 나옵니다 — 이슈·코멘트·전체 이력·첨부·
-위키 페이지가 동기화된 워크스페이스에서 standalone 워크스페이스로 옮겨지고,
+위키 페이지가 동기화된 워크스페이스에서 gadak 이 직접 보관하는 워크스페이스로
+옮겨지고,
 끝에 원본 대 이전본 건수 대조표가 출력됩니다:
 
 ```bash
@@ -216,10 +218,10 @@ macOS 앱에서는 같은 설치가 버튼이기도 합니다. **설정 → 연
 
 ## 무엇을 커버하나
 
-Connected는 Atlassian Cloud와 대화합니다. Standalone(0.16부터)은 Atlassian
+Connected는 Atlassian Cloud와 대화합니다. gadak 자신의 트래커(0.16부터)는 Atlassian
 계정이 없는 워크스페이스입니다. 그 origin은 앱과 함께 다니는 미니멀한
 Jira입니다. 어느 쪽이든 미러는 캐시이고, 모든 쓰기는 origin을 통과합니다.
-Standalone에서 영속 파일은 origin의 persist 파일, 즉 워크스페이스 origin
+gadak origin 에서 영속 파일은 origin의 persist 파일, 즉 워크스페이스 origin
 폴더의 `issuetap.db`(SQLite, WAL)입니다. gadak이 꺼져 있을 때 복사하거나
 (`-wal`/`-shm` 사이드카 포함), `sqlite3 origin/issuetap.db ".backup
 dest.db"`를 쓰세요.
@@ -231,7 +233,7 @@ Jira 알림함은 안 됩니다. 그 일은 Jira에 남습니다.
 <details>
 <summary>▶ 전체 매트릭스와 ✅마다 붙은 각주</summary>
 
-| | Connected (Atlassian Cloud) | Standalone (0.16부터) |
+| | Connected (Atlassian Cloud) | gadak 자신의 트래커 (0.16부터) |
 | --- | :---: | :---: |
 | 이슈 읽기·검색 (FTS, JQL, SQL) | ✅¹ | ✅¹ |
 | 생성, 코멘트, 상태 전이, 담당자, 라벨, 우선순위 | ✅ | ✅ |
@@ -423,7 +425,7 @@ JQL로 여전히 물을 수 없는 것은 `gadak sql`과
 ## 설치
 
 brew 두 줄은 이 페이지 맨 위에 있습니다. Atlassian Cloud에 연결하거나,
-(0.16부터) Atlassian 계정이 필요 없는 standalone 워크스페이스로 시작합니다.
+(0.16부터) Atlassian 계정이 필요 없는 gadak origin 워크스페이스로 시작합니다.
 connected 사이트는
 [API 토큰](https://id.atlassian.com/manage-profile/security/api-tokens) 하나면
 되고, 그 토큰이 같은 사이트의 Jira와 Confluence를 함께 커버합니다.

README.md

@@ -35,7 +35,7 @@ stays the source of truth.
 </p>
 
 A connected site needs one [API token](https://id.atlassian.com/manage-profile/security/api-tokens)
-— it covers Jira and Confluence on the same site. A standalone workspace
+— it covers Jira and Confluence on the same site. A workspace whose origin is gadak's own tracker
 needs no Atlassian account at all.
 
 **You pick what it mirrors.** The wiki is off until you ask for it, and when
@@ -77,7 +77,7 @@ gadak serve
 ```
 
 Or leave one, data in hand — issues, comments, full history, attachments,
-and wiki pages move from a synced workspace into a standalone one, ending
+and wiki pages move from a synced workspace into a gadak-origin one, ending
 with a source-vs-migrated count table:
 
 ```bash
@@ -210,10 +210,10 @@ open-by-key half.
 
 ## What's covered
 
-Connected talks to Atlassian Cloud. Standalone (from 0.16) is a workspace
+Connected talks to Atlassian Cloud. gadak's own tracker (from 0.16) is a workspace
 with no Atlassian account — a minimal Jira origin that travels with the
 app. The mirror is a cache either way; every write goes through the origin.
-On standalone the durable file is the origin's persist file — issuetap.db
+On a gadak origin the durable file is the origin's persist file — issuetap.db
 in the workspace's origin folder (SQLite, WAL). Copy it while gadak is not
 running (include the `-wal`/`-shm` sidecars), or
 `sqlite3 origin/issuetap.db ".backup dest.db"`.
@@ -225,7 +225,7 @@ dashboards and Jira's notification inbox do not — those stay in Jira.
 <details>
 <summary>▶ The full matrix, with the footnote for every ✅</summary>
 
-| | Connected (Atlassian Cloud) | Standalone (from 0.16) |
+| | Connected (Atlassian Cloud) | gadak's own tracker (from 0.16) |
 | --- | :---: | :---: |
 | Issue read and search (FTS, JQL, SQL) | ✅¹ | ✅¹ |
 | Create, comment, transition, assignee, labels, priority | ✅ | ✅ |
@@ -416,7 +416,7 @@ and its off switch.
 ## Install
 
 The two brew lines are at the top of this page. Atlassian Cloud, or (from
-0.16) a standalone workspace with no Atlassian account — a connected site
+0.16) a workspace whose origin is gadak's own tracker with no Atlassian account — a connected site
 needs one [API token](https://id.atlassian.com/manage-profile/security/api-tokens),
 which covers Jira and Confluence on the same site.
 

SECURITY.md

@@ -181,7 +181,7 @@ device. A DNS-named Host — the shape `tailscale serve` forwards — is
 rejected by the rebinding guard unless the request is bound for a surface a
 token authenticates, and there are exactly three of those:
 
-- **The origin passthrough** (`/api/v1/origin`, standalone workspaces): raw
+- **The origin passthrough** (`/api/v1/origin`, gadak-origin workspaces): raw
   REST for paired gadak machines and this machine's own routed writes
   (`internal/server/origin_rest.go`).
 - **The mirror REST** (any workspace kind): everything the local web UI
@@ -374,12 +374,12 @@ Offboarding depends on what the profile holds. On a connected workspace —
 or a paired one — the origin is elsewhere (your Atlassian site, or the home
 serve you paired with), and `rm -rf ~/.gadak` is the whole story: it removes
 the mirror, the credential, and every profile, and nothing else on the
-machine or in Jira knows gadak existed. On a standalone workspace that
+machine or in Jira knows gadak existed. On a workspace whose origin is gadak's own tracker that
 command destroys the origin itself: each profile's `origin/issuetap.db` is
 the only copy of that tracker anywhere. Copy the file out first (plain
 SQLite — nothing of gadak's is needed to read it) unless you mean to throw
 the data away. To offboard one profile rather than the whole home,
-`gadak workspaces rm <name>` removes just that profile; a standalone one
+`gadak workspaces rm <name>` removes just that profile; a gadak-origin one
 additionally requires `--destroy-origin`, for exactly the reason above.
 
 ## Release artifacts

contrib/omarchy/README.md

@@ -5,7 +5,7 @@ that answers one question offline: how much of the local gadak mirror is
 open, and how much of that has been sitting in the same status for more
 than seven days.
 
-This is not a “my assigned tickets” widget. A standalone workspace has no
+This is not a “my assigned tickets” widget. A workspace whose origin is gadak's own tracker has no
 account (`gadak init --local` clears `AccountID`), and the competing
 community plugins (`tmn73.jira`, Linear, Todoist, ClickUp, `37signals.basecamp`)
 all need a cloud token and a live network. This one runs `gadak sql --json`

docs/AGENT_SETUP.md

@@ -40,7 +40,7 @@ machine already uses. Steps:
 ## No Jira on this machine
 
 If there is no Atlassian account and the user wants a backlog (this repo,
-this machine, an agent-owned plan), that is a **standalone** workspace — not
+this machine, an agent-owned plan), that is a workspace with a **gadak origin** — not
 a missing Jira token. Do not invent `TODO.md` or a GitHub Issue when `gadak`
 is on PATH. If this machine already has a connected gadak workspace, use a
 dedicated `--workspace` so personal issues never land on the company site.
@@ -49,7 +49,7 @@ Paste this prompt:
 
 ```text
 This machine has no Jira account. Keep a backlog for the work we do: create
-a gadak standalone workspace, file the first tickets, and show them to me.
+a gadak gadak-origin workspace, file the first tickets, and show them to me.
 
 1. Install gadak if missing: `brew install midagedev/tap/gadak-cli`
    (or the install script in the repo README).
@@ -99,7 +99,7 @@ Jira issues are mirrored to a local SQLite file. Prefer these over any Jira API:
   the Jira-site escape hatch; `gadak views open` is open-in-gadak.
 - `gadak comment <KEY> -m "…"`, `gadak transition <KEY> "<status>"` — writes go
   through the origin (Jira on a connected workspace, the local origin on a
-  standalone one).
+  gadak-origin one).
 
 Rules: filter on `status_category` ('new'|'inprogress'|'done') and ids, never
 on display names — Jira localizes those per account. Query the `issues_full`

docs/ARCHITECTURE.md

@@ -40,7 +40,7 @@ flowchart TB
 Confluence is a peer source (`internal/confluence`, `decisions/0006-confluence-connector.md`):
 page writes (create, edit, page comments — [GDK-380]/381/382) go through
 `origin.Wiki`, never the SQLite mirror. On a connected workspace that client
-is Confluence REST; on a standalone workspace it is the in-process issuetap
+is Confluence REST; on a workspace whose origin is gadak's own tracker it is the in-process issuetap
 handler (same session as `origin.Client`), whose Confluence API receives the
 same verbs.
 

docs/CONCEPT.md

@@ -42,7 +42,7 @@ product than this one.
 **The mirror is disposable.** If the schema is a cache of the origin's truth,
 deleting it is always safe. That property is worth protecting: the moment the
 mirror holds something irreplaceable it becomes infrastructure someone has to
-back up. On standalone, the irreplaceable file is not `gadak.db` — it is the
+back up. On a gadak origin, the irreplaceable file is not `gadak.db` — it is the
 origin persist file (below).
 
 ## Two origins
@@ -52,7 +52,7 @@ A workspace is bound to one origin.
 - **Connected** — Atlassian Cloud (or Linear, when configured) is the record.
   The token talks to that site. Delete the profile directory and you have lost
   a cache.
-- **Standalone** (from 0.16) — the origin is an in-process minimal Jira
+- **A gadak origin** (from 0.16) — the origin is an in-process minimal Jira
   (`issuetap`). There is no Atlassian account. The durable file is
   `origin/issuetap.db` in the profile directory (`internal/origin/origin.go`
   `PersistRel`): a SQLite database (WAL). That is the file to back up —
@@ -132,7 +132,7 @@ CLI covers those plus `create`, `attach`, and `edit` (values always come from
 what the origin allows, never free text — an issue type or priority the CLI
 cannot match is refused with the names that origin actually uses).
 Wiki page create, edit, and comment go through the origin too — Confluence
-Cloud on a connected workspace, the in-process origin on standalone.
+Cloud on a connected workspace, the in-process origin on a gadak origin.
 
 Hierarchy is first-class: `epic_key` is derived honestly (the nearest epic
 *ancestor*, so a sub-task groups under its epic, not its story), group-by-epic
@@ -175,7 +175,7 @@ Not a Jira replacement — we do not reimplement boards, dashboards, or the rest
 of Jira's UI; we contain those pages so the window can hold what the mirror
 refuses to model. Not a sync engine. Not an archive. Not multi-user. On a
 connected workspace, not a place to put anything you cannot afford to lose
-(the Atlassian site holds the record). On standalone, `origin/issuetap.db`
+(the Atlassian site holds the record). On a gadak origin, `origin/issuetap.db`
 *is* the record — losing that file loses the work.
 
 ## Good fit / bad fit

docs/CONFIGURATION.md

@@ -459,9 +459,9 @@ is nothing to say, and a client-supplied value is ignored.
 | `confluence` | object or absent | absent = wiki mirror off. `gadak init --local` writes the block scoped to `LOC` | Settings → Sources / `gadak config set confluence` | Next Confluence pass |
 | `linear` | object or absent | absent = Linear source off. `apiKey` (personal API key, sent bare in the Authorization header) turns the source on; writes to Linear-owned keys route through it | edit `config.json` (no Settings surface yet) | Next `sync --source linear` |
 | `linear.teamIds` | string[] | `[]` = every team the key can see; team UUIDs restrict the mirror scope | edit `config.json` | Next Linear pass |
-| `devStatus` | bool | **false** | `gadak config set devStatus true` / `config.json` (not on Settings UI or Settings PUT) | Next sync; connected Cloud: mirror Jira's development-status API into `dev_links` (one extra request per issue). Standalone always fetches; `gadak dev link` / `dev scan` write the same table |
-| `actor` | object or absent | absent = no acting identity; env `GADAK_ACTOR` (`slug\|display name`) overrides, and Claude Code sessions are auto-detected when both are unset | `gadak config set actor 'slug\|display name'` / `config.json` (not on Settings UI or Settings PUT; never team-exported) | Next origin session; writes to an issuetap origin (standalone/paired) carry `X-Issuetap-Actor` and attribute to that agent account. Never sent to connected Jira/Linear |
-| `locale` | string | _(empty)_ = English; `en` \| `ko` \| `ja` \| `de` | `gadak config set locale ko` / `config.json` (not on Settings UI or Settings PUT) | Standalone only: the origin's display-name language — status / issue-type / field names and agent aliases follow it; priority names stay English, like a live Cloud site. Changing it rebuilds the mirror on the next sync (display names are cached). A connected workspace ignores it: its language is the Atlassian account's |
+| `devStatus` | bool | **false** | `gadak config set devStatus true` / `config.json` (not on Settings UI or Settings PUT) | Next sync; connected Cloud: mirror Jira's development-status API into `dev_links` (one extra request per issue). A gadak origin always fetches; `gadak dev link` / `dev scan` write the same table |
+| `actor` | object or absent | absent = no acting identity; env `GADAK_ACTOR` (`slug\|display name`) overrides, and Claude Code sessions are auto-detected when both are unset | `gadak config set actor 'slug\|display name'` / `config.json` (not on Settings UI or Settings PUT; never team-exported) | Next origin session; writes to a gadak origin (local or paired) carry `X-Issuetap-Actor` and attribute to that agent account. Never sent to connected Jira/Linear |
+| `locale` | string | _(empty)_ = English; `en` \| `ko` \| `ja` \| `de` | `gadak config set locale ko` / `config.json` (not on Settings UI or Settings PUT) | A gadak origin only: the origin's display-name language — status / issue-type / field names and agent aliases follow it; priority names stay English, like a live Cloud site. Changing it rebuilds the mirror on the next sync (display names are cached). A connected workspace ignores it: its language is the Atlassian account's |
 | `confluence.spaces` | string[] | `[]` = every *global* space; personal spaces only if named (`internal/config/config.go`) | Settings → Sources / `gadak config set confluence.spaces` | Next Confluence pass |
 | `terminal` | `{shell, workingDir, scrollback, cursorBlink}` | absent = all defaults (see below) | `gadak config set terminal` or `terminal.<leaf>` (not on Settings UI or Settings PUT; never team-exported — shell and workingDir are this machine's paths) | Next terminal session create; a block set replaces the whole object, a leaf set merges |
 
@@ -652,7 +652,7 @@ into `fields`.
 watches, favorites, and recents (`cmd/gadak/export.go`; help in
 `cmd/gadak/help.go`). Credentials never appear in the file — a credential-shaped
 string is refused (`secretscan`). It is not a `gadak team export` file (team
-settings live in that other command) and it does not include the standalone
+settings live in that other command) and it does not include the gadak-origin
 persist file (`origin/issuetap.db`) or the issue rows in `gadak.db`.
 
 `gadak import <FILE>` restores those four lists. On a name/key conflict the
@@ -717,8 +717,8 @@ against `/myself` before anything is written.
 | --- | --- |
 | `$GADAK_HOME/config.json` or `~/.gadak/config.json` | Settings + credential (0600) |
 | `$GADAK_HOME/gadak.db` | SQLite mirror (a cache; the next sync rebuilds it from the origin) |
-| `$GADAK_HOME/origin/issuetap.db` | Standalone origin persist (`internal/origin/origin.go` `PersistRel`). SQLite (WAL); this is the record on a standalone workspace. Copy while gadak is not running (include `-wal`/`-shm`), or `sqlite3 <db> ".backup"`. Absent on a connected workspace. A sibling `origin/issuetap.yaml` is a one-shot seed if the db is missing. |
-| `~/.gadak/profiles/<name>/` | Isolated config + mirror (and, on standalone, persist) per profile |
+| `$GADAK_HOME/origin/issuetap.db` | A gadak origin's persist file (`internal/origin/origin.go` `PersistRel`). SQLite (WAL); this is the record on a workspace whose origin is gadak's own tracker. Copy while gadak is not running (include `-wal`/`-shm`), or `sqlite3 <db> ".backup"`. Absent on a connected workspace. A sibling `origin/issuetap.yaml` is a one-shot seed if the db is missing. |
+| `~/.gadak/profiles/<name>/` | Isolated config + mirror (and, on a gadak origin, persist) per profile |
 
 Never write issue rows into the DB by hand — the next sync overwrites them. The
 supported external write table is `enrichments` (see [PLUGINS.md](PLUGINS.md)).