ghosty-acp 0.0.76 → 0.0.78

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "ghosty-acp",
3
- "version": "0.0.76",
3
+ "version": "0.0.78",
4
4
  "description": "Conecta tu editor a un agente ACP remoto. Puente entre entrada/salida estándar y WebSocket, sin dependencias.",
5
5
  "type": "module",
6
6
  "bin": {
@@ -5,7 +5,7 @@ license: MIT
5
5
  compatibility: Needs Node 22+ (for npx @ghostystudio/cli) or curl, and network access to https://www.ghosty.studio
6
6
  metadata:
7
7
  author: ghosty-studio
8
- version: "1.109"
8
+ version: "1.115"
9
9
  ---
10
10
 
11
11
  # Configure a Ghosty Studio agent
@@ -109,7 +109,7 @@ legacy and always `true`.)
109
109
  | "something in Ghosty is broken and I can't fix it", "talk to Ghosty / support" | `ghosty support open --title "<short>" --body "<what, when, what was tried>"` — support answers BY EMAIL; replying to that email adds to the case. `ghosty support ls` / `show <id>` / `reply <id> --body …` to follow up. The case carries the account and agents' recent errors; never paste secrets |
110
110
  | "archive / pause this agent for good", "I don't use this agent anymore" | prefer archiving over deleting: `ghosty agents archive <id>` (stops answering everywhere, hidden, nothing deleted; deleted automatically after 7 days). `ghosty agents restore <id>` brings it back as it was; `ghosty agents ls --archived` lists them |
111
111
  | "rotate / revoke the agent token", "the gat_ token leaked" | owner session only (a `gat_` token can't rotate itself): `ghosty agents token rotate <id> --yes` prints the new token ONCE on stdout — hand it to the user to store, never echo it into logs; `token revoke <id> --yes` closes the editor channel. On ACP engines it cuts the turn in flight |
112
- | "teach it a rule for everyone", "from now on don't offer X", "add / turn off a promo" | `ghosty agents memory ls <id> --json`; rule: `memory add <id> "…" [--scope wa]` (travels in every conversation, also with customers), `memory edit <id> <rule-id> "…"`, `memory rm <id> <id>`; promo: `memory add <id> --promo --title T --details D [--from YYYY-MM-DD] [--until YYYY-MM-DD]`, `memory enable|disable <id> <promo-id>`. Personal preferences of the user go in `me memories` instead |
112
+ | "teach it a rule for everyone", "from now on don't offer X", "add / turn off a promo" | `ghosty agents memory ls <id> --json` (a rule with `delivered: false` no longer fits and does not reach the agent: tell the user to move old rules to the prompt from the agent page → Memory); rule: `memory add <id> "…" [--scope wa]` (travels in every conversation, also with customers), `memory edit <id> <rule-id> "…"`, `memory rm <id> <id>`; promo: `memory add <id> --promo --title T --details D [--from YYYY-MM-DD] [--until YYYY-MM-DD]`, `memory enable|disable <id> <promo-id>`. Personal preferences of the user go in `me memories` instead |
113
113
  | "give / hand over this agent to someone", "move it to another workspace" | `ghosty agents transfer <id> --to EMAIL [--workspace THEIR-SLUG] --dry-run`, show the plan and warnings (missing provider key), confirm with the user, then without `--dry-run` + `--yes`. Between the user's own workspaces: `--workspace SLUG|none` (none = personal; `--force` only if they accept cutting a running turn). Owner session, never the `gat_` token |
114
114
  | "put my agent in Microsoft Teams" (MTeams; not Ghosty Teams) | the company creates a Single Tenant **Azure Bot** (F0) in its own Azure and adds the Teams channel; then `ghosty channels mteams connect <id> --app-id <guid> --tenant <guid> [--name N]` with the client secret in `GHOSTY_MTEAMS_SECRET` (Ghosty tests it with Microsoft); paste the returned messaging endpoint in the Azure Bot → Configuration; `ghosty channels mteams zip <id> <bot> --out F.zip` and the Teams admin uploads it in Manage apps. No app store. The admin can try it first with Azure's “Test in Web Chat”. If it stays silent, `ghosty channels mteams ls <id>` shows the last problem (`lastError`: e.g. "el token no es para este bot" = endpoint pasted into another Azure Bot). Docs: /docs/canales/microsoft-teams |
115
115
  | "connect WhatsApp Business / Messenger" | `ghosty channels whatsapp connect <id> --json` starts or resumes the connection and gives `url`: pass it to the user, it works for 1 h on any screen (if the phone doesn't finish Meta's sign-in, open it on a computer). Run it again to see `state`: `pending` or `connected`. It needs no balance and doesn't turn the agent on. Messenger: `messenger connect` opens Channels |
@@ -139,17 +139,26 @@ legacy and always `true`.)
139
139
  | "clean up its history / delete the test chats" | `ghosty conversations rm <id> --kind test --before 7d --dry-run`, confirm, then without `--dry-run` (`--yes`); customer threads: `conversations archive <id> --kind channel` (never rm); only the user's own chats with it: `conversations rm <id> --kind personal --dry-run` (also gone from their apps) |
140
140
  | "merge my duplicated documents" | `ghosty agents docs dedupe --dry-run`, show the merges, then `--yes`; two specific ones: `ghosty agents docs dedupe <keep-id> <drop-id>` |
141
141
  | "use this file from my files in the chat" | `ghosty me files ls --q <name> --json` → id, then `ghosty chat <id> "…" --file <file-id>` |
142
+ | "why isn't the bot answering this card / what is the board's fixed hour" | `ghosty boards pauses <workspace> [board] --json`: bot on, paused until, only operator, phone echo; pause a card with `ghosty cards pause <agent-id> <card> --for 90m | --until 17:30 | --mode board_until|permanent`, resume with `cards resume`; set the hour with `ghosty boards set-hour <workspace> [board] 17:30|off` |
143
+ | "how many times did the agent escalate to a person and why" | `ghosty boards escalations <workspace> [board] --since 7d --json` (logged since 2026-11-20) |
144
+ | "did a customer wait for a box / who got no reply (plan box cap)" | `ghosty spaces waits <workspace> --since 7d --json` (`gave_up` = no reply); caps: `ghosty usage --boxes --workspace W` |
145
+ | "the owner asked to remove their agency (partner)" | `ghosty spaces partners <workspace>` lists paused partners; `--restore <id>` (owner or staff) or `--confirm <id>` (staff only) |
146
+ | "what does this workspace have / are seats, agents or boxes at the cap" | `ghosty spaces show <workspace> --json`; agents with signals: `ghosty agents ls --workspace W --health`; staff, boxes at cap: `ghosty usage --boxes --at-cap` |
147
+ | "change someone's role / name / expiry, resend the invite, give campaigns" | `ghosty spaces members role|name|resend|until|marketer-add|marketer-rm <workspace> <email> [--role member|admin] [--name TEXT] [--until DATE|none]`; staff access: `members rm <workspace> <email> --support` |
148
+ | "who owns this card / archive one card / hide a board view" | `ghosty cards assign <agent-id> <card> --to <email>|none`, `ghosty cards archive <agent-id> --card <card>`, `ghosty boards views <workspace> [board] --hide tabla` |
149
+ | "which connectors does this agent really have" | `ghosty agents connectors <agent-id>` (granted vs connected vs no effect) |
142
150
  | "which workspaces do I have / where is X's workspace" | `ghosty spaces ls [--user <email>] --json` (others: staff) |
143
151
  | "no claude model fits / raise the model cap" (422 «caps at 0.25×») | staff: `ghosty spaces set <workspace> --model-cap 4 --dry-run`, then `--yes` (1 = up to Sonnet, none = its plan's) |
144
152
  | "create a workspace" | `ghosty spaces create <slug> --dry-run`, then without `--dry-run` (yours, trial; same rules as the web: confirmed email, plan cap). `--combo` other than teams: staff |
145
- | "the agent should only answer new people on my number / don't let it answer my mom" | `ghosty channels whatsapp contacts <id> --integration <id>` (state: new-only, known contacts, muted) · `--new-only on` (skips people the owner already talks to: WhatsApp contacts + history + people the owner wrote first) · `--mute <phone>` (agent never answers that number) · `--allow <phone>` (agent always answers it, even if known) · `--unmute <phone>` (back to the general rule) · `--signature 👾|off` (prefix on the agent's messages so customers tell it from the owner; new numbers start with 👾) · `--seed` (bring known contacts again) · `--list [--q mom]` (the phone contacts with names and each rule; find a contact by name before muting it). Numbers connected from Channels start not answering, with new-only on: turn the number on once the known contacts show up |
153
+ | "the agent should only answer new people on my number / don't let it answer my mom" | `ghosty channels whatsapp contacts <id> --integration <id>` (state: new-only, known contacts, muted) · `--new-only on` (skips people the owner already talks to: WhatsApp contacts + history + people the owner wrote first) · `--mute <phone>` (agent never answers that number) · `--allow <phone>` (agent always answers it, even if known) · `--team <phone>` (that number is the owner's team, e.g. their other business line: the agent answers it in team mode, with no sales script, no lead card and no campaigns) · `--unmute <phone>` (back to the general rule) · `--signature 👾|off` (prefix on the agent's messages so customers tell it from the owner; new numbers start with 👾) · `--seed` (bring known contacts again) · `--list [--q mom]` (the phone contacts with names and each rule; find a contact by name before muting it). Numbers connected from Channels start not answering, with new-only on: turn the number on once the known contacts show up |
146
154
  | "bring only the ad leads from this number" | `ghosty channels whatsapp history <id> --integration <formmy-id> --dry-run --list --only-ads` → show which, then without `--dry-run` (`--yes`); new numbers: `channels whatsapp link … --ads-only on` |
147
155
  | "this agent answers two numbers, keep their boards separate" | `ghosty channels whatsapp link <id> --integration <formmy-id> --board "<board>" --dry-run` per number (`--board agent` = back to the agent's); `channels whatsapp ls <id>` shows each number's board |
148
- | "why did it fail? / it didn't answer" | `ghosty turns ls <id> --errors --since 24h --json` → failed turns with their `error`; a WhatsApp customer got NO turn at all: `ghosty turns ls <id> --skipped --since 24h --json` → `reason` (manual_mode, records_only, paused, reaction, channel_off, no_quota, own_number) |
156
+ | "why did it fail? / it didn't answer" | `ghosty turns ls <id> --errors --since 24h --json` → failed turns with their `error`; a WhatsApp customer got NO turn at all: `ghosty turns ls <id> --skipped --since 24h --json` → `reason` (manual_mode, records_only, paused, reaction, channel_off, no_quota, own_number, loop_brake = more than 30 messages from one contact in 10 min or 60 in 1 h, usually another bot; paused 2 h) |
149
157
  | "what did it answer to that? / audit a reply" | `ghosty turns show <id> [turn-id] --json` → `input` (what came in) and `output` (what it answered), any engine or channel |
150
158
  | "where did it say X? / find the turn where it quoted…" | `ghosty turns grep <id> "text" --since 7d [--role user\|agent] --json` → `matches[]` with `turnId` and `snippet`; then `turns show <id> <turnId>`. Text only, not tools |
151
- | "set the quote issuer / VAT / folio prefix" (house `cotizar` tool) | `ghosty agents set <id> --quote iva=0,prefix=SFI,vigencia=15` (empty value clears: `rfc=`); `ghosty agents get <id> --fields cotizacion` to read it. An own skill named `cotizar` already turns the house tool off |
159
+ | "set the quote issuer / VAT / folio prefix" (house `quote_create` tool) | `ghosty agents set <id> --quote iva=0,prefix=SFI,vigencia=15` (empty value clears: `rfc=`); `ghosty agents get <id> --fields cotizacion` to read it. An own skill named `cotizar` already turns the house tool off |
152
160
  | "let the owner edit catalog prices themselves" (web, CSV or chatting with the agent) | Put the catalog in a database: `ghosty dbs create <id> catalogo` → `ghosty dbs import <id> catalogo --from csv:productos.csv --table productos` → `ghosty agents set <id> --quote catalogo_base=catalogo:productos`. Columns are detected by name (sku/clave/codigo, nombre, precio, unidad, categoria). If the database does not answer, the agent keeps quoting with the last good read or the `--catalog` list |
161
+ | "answer with the products, prices and photos of my online store" (WooCommerce or Shopify) | `ghosty dbs create <id> catalogo-tienda --read` → `ghosty dbs import <id> catalogo-tienda --from store:https://myshop.com --dry-run`, then without `--dry-run`. Fills the `productos` table (name, price range, description, sale, link, permanent photo) and the `presentaciones` table (one row per size or variant with its own price). The agent finds products with `product_search` (accent- and typo-tolerant, returns the current price and available sizes; works on any table with a name and a price column) and sends each product as a photo with its caption using `product_send`; what the store won't sell ends up `disponible = 0` and is never sent. Run the import again to refresh prices. |
153
162
  | "prices already include VAT except bulk / VAT only with invoice" | `ghosty agents set <id> --catalog productos.csv` (sku,nombre,precio,unidad,categoria) + `--quote iva-incluido=true,iva.GRANEL=16,solo-factura=true`; the category comes from the catalog, not the model; the agent passes `factura: true` when the customer asks for an invoice |
154
163
  | "why was it slow? / where did the time go?" | `ghosty turns show <id> [turn-id] --json` → `byTool` (seconds per tool), `steps` (timeline), `totals.outputTokens`; pool engines only (`timeline: false` otherwise) |
155
164
  | "is it running the new image? / it still behaves old" | `ghosty agents box <id> --check <path-the-new-image-brings> --json` → `stale`; if > 0, `ghosty agents box <id> --recycle`, one turn, check again (ACP: `agents restart`) |
@@ -176,6 +185,7 @@ legacy and always `true`.)
176
185
  | "what did this agent run / did it leak anything" (support) | staff only: `ghosty conversations show <agent> <conversation> --tools` — every tool with its command (secrets masked) and the external domains it named; never while a turn runs there. Who sees a thread: `--why`; someone's threads: `ghosty conversations ls --user <email>` |
177
186
  | "are images being metered / how much do images cost on Free" | staff only: `ghosty usage --images --plan free --since 7d [--by user]` — charges vs images agents delivered (far more deliveries than charges = the meter isn't reporting) and stored cost vs the table |
178
187
  | "did the push reach them / did the announcement go out" | staff only: `ghosty push log --user <email> --since 2d` (announcements show as `novedad`); `ghosty novedades ls` shows delivered/rejected per announcement |
188
+ | "what did my phone notify me / what is still waiting for me" | `ghosty notifications [--pending]` (= the app's «Avisos» bell; urgent sales ones stay pending until someone replies to that customer) |
179
189
  | "my site is paused / will pause", "keep my page online", "pin my site", "reactivate my page" | `ghosty sites ls` (state: pinned, paused, pauses on…), then `ghosty sites pin <slug>` — «Sitios fijos»: $49 USD/yr or $5 USD/mo ($990 / $99 MXN), includes up to 3 sites; never pauses, no banner. With a free slot it pins at once; otherwise it prints a card-payment link to hand over as is; `needs_pack` = all slots used: ask, then `--add-pack`. `--interval month`, `--currency mxn`. `ghosty sites unpin <slug>` frees a slot; `ghosty sites pins [--cancel]`. "delete my site": confirm first, then `ghosty sites rm <slug> --yes` (can't be undone) |
180
190
  | "how is my app", "is it awake", "what variables / tasks / files does it have" | `ghosty sites show <slug> --json` (status `site.wake.phase`: ready/starting/asleep/failed, never wakes it; variable NAMES only). Change: `ghosty sites env set <slug> NAME=value` · `env rm <slug> NAME`; tasks `ghosty sites cron ls|pause|resume|rm <slug> [id]`; files `ghosty sites files ls|rm <slug> [id]`. Errors (server + browser, grouped): `ghosty sites errors <slug> [--since 6h] [--req-id ID]`. The agent inside the app: `ghosty sites agent <slug>` (show) · `sites agent <slug> create` (Haiku 5.5) · `--agent <id>` / `--model m`. The person sees the same at `/c/sitios/<slug>` (tabs `?tab=datos` etc.) |
181
191
  | "put my domain on my app", "use reservas.midominio.com", "custom domain" | `ghosty sites domain add <slug> <domain>` prints the DNS record (CNAME to `<slug>.ghosty.page` for a subdomain, A to the printed IP for the root); hand it over in one line. After they add it: `ghosty sites domain verify <slug> <domain>` (DNS can take minutes to a couple of hours; Cloudflare proxy off). Verified = the app answers at `https://<domain>`, HTTPS issued on the first visit. Apps with a backend only. `ghosty sites domain ls <slug>`, `rm <slug> <domain>` |
@@ -213,6 +223,7 @@ legacy and always `true`.)
213
223
  | "what do my agents remember about me?" / "make it remember X" | `ghosty me memories ls --json`; `me memories add "X"` (app-wide, all agents), `edit <id> "…"`, `rm <id>` (only travels in personal chats) |
214
224
  | "the doc the agent wrote" (a `/app/docs/<id>` link): read it, PDF, check spelling, read it aloud, go back a version | `ghosty me docs ls --json`; `me docs get <doc|link> --json` (blocks + versions); `me docs export <doc> --format md` (stdout) or `--format pdf --out F.pdf`; `me docs check <doc> [--language en-US] --json`; `me docs tts <doc> --block N [--segment S] --out F.wav`; `me docs restore <doc> <n>`; `--rev N` for an older version. Not the same as `agents docs` (files an agent reads) |
215
225
  | "pause / resume this conversation", "stop the bot with this customer", "take it out of operator-only" | `ghosty cards pause <id> <folio\|phone> --mode 30min\|2h\|24h\|until_tomorrow\|permanent` (permanent = escalated, operator only) / `ghosty cards resume <id> <card>` (also lifts operator-only; it does NOT answer by itself). Both are logged |
226
+ | "write to this card as the agent", "send this template/text to #147", "react 🫡 to their message" (no paid turn) | `ghosty cards send <id> "#147" --text "…" \| --template NAME --var a --var b \| --react 🫡`; try `--dry-run` first (shows the 24 h window and approved templates); it messages a real customer, so confirm or pass `--yes` |
216
227
  | "answer the ones left pending", "detonate pending replies", "ask the bot to reply to this customer" | list first: `ghosty cards pending <id> [--since 2026-10-06T02:13Z\|12h] --json` → `cards[].flags` (paused, escalated, window_closed, reaction, not_whatsapp, no_agent: those are NEVER answered). Show the user how many would be answered (`--reply --dry-run`), and only after they confirm: `ghosty cards pending <id> --since … --reply --yes` (real customers get messages). One card: `ghosty cards reply <id> <card> [--directive "…"] --yes`. Never on a production conversation without the user's explicit OK |
217
228
  | "who was answering this number?", "which numbers/pages were left without an agent?" | `ghosty channels whatsapp integrations <id> --json` → `lastAnswer {agentName, at}` per number; `--orphans` lists only numbers nobody answers today that had turns (`channels messenger ls <id> --orphans` for pages). Give it back with `ghosty channels whatsapp enable <that-agent> <number> --dry-run`, then `--yes` after the user confirms |
218
229
  | "change this paragraph / section of the prompt", "patch the prompt", "add this to the prompt" | write `{"edits":[{"what":"hours","find":"9 to 6","replace":"9 to 7"}]}` to a file and run `ghosty agents set <id> --prompt-patch edits.json --dry-run` (prints the diff), then without `--dry-run`. Append: `--prompt-append extra.md`. Applies on the server's CURRENT prompt (never `get --prompt-out` + `set --prompt-file` for a small change: it overwrites concurrent edits). An edit that doesn't find its place → 409, nothing changes |
@@ -16,9 +16,9 @@ curl -s "$B" -H "Authorization: Bearer $GHOSTY_AGENT_TOKEN"
16
16
 
17
17
  ## PATCH /
18
18
  Body: any of `{ "name", "model", "prompt", "webSearch": bool, "channels": { "teams": bool, "web": bool, "whatsapp": bool, "messenger": bool }, "starters": ["…up to 5, ≤80 chars"], "tools": { "gs": { "<name>": bool }, "extensions": { "<name>": bool } } }`.
19
- Also `"channelRoles": { "teams": "equipo"|"cliente", "programado": … }` (who talks on that channel; the rest are fixed) and `"email": { "anyFromClients": bool }` (owner only: `enviar_correo` to any address from customer channels). CLI: `agents set --web-search --chat-starters --channel-roles --extensions --email-any`.
20
- `tools.gs` today: `programar_seguimiento` (schedule a future turn), `archivo_enviar` (send a file the agent produced — an https URL — to the person it is talking to: photo/audio/document on WhatsApp and Messenger, a link elsewhere) and `actualizar_identidad` (the agent rewrites or appends to its own prompt — it only works when the owner talks to it from Studio or the Mac app; from Messenger/WhatsApp/Teams the call is refused).
21
- `tools.canales` decides per channel: `{ "canales": { "messenger": { "actualizar_identidad": false } } }` (channels: `chat` = Studio/Mac app, `teams`, `whatsapp`, `messenger`, `programado`, `prueba`; the GET returns the effective matrix). By default `actualizar_identidad` is only on in `chat`.
19
+ Also `"channelRoles": { "teams": "equipo"|"cliente", "programado": … }` (who talks on that channel; the rest are fixed) and `"email": { "anyFromClients": bool }` (owner only: `email_send` to any address from customer channels). Also `"fallback": bool` (default true: if the owner's Claude key or subscription fails before answering, the turn is retried once with Haiku on Ghosty's key and billed to the plan). CLI: `agents set --web-search --fallback --chat-starters --channel-roles --extensions --email-any`.
20
+ `tools.gs` today: `followup_schedule` (schedule a future turn), `file_send` (send a file the agent produced — an https URL — to the person it is talking to: photo/audio/document on WhatsApp and Messenger, a link elsewhere) and `identity_update` (the agent rewrites or appends to its own prompt — it only works when the owner talks to it from Studio or the Mac app; from Messenger/WhatsApp/Teams the call is refused).
21
+ `tools.canales` decides per channel: `{ "canales": { "messenger": { "identity_update": false } } }` (channels: `chat` = Studio/Mac app, `teams`, `whatsapp`, `messenger`, `programado`, `prueba`; the GET returns the effective matrix). By default `identity_update` is only on in `chat`.
22
22
  `tools` turns tools off (`false`) or back on (`true`) by name and merges with what is saved. `tools.gs` are the tools gs lends the agent (the GET lists them all with their state); `tools.extensions` are the machine's `config.yaml` extensions (`developer` = shell + files, `todo`, `analyze`, `easybits`…; `?full=1` returns the real list as `extensions`) and changing them restarts the machine. The veto is enforced: an off tool disappears from the list and is refused when called.
23
23
  Response: the same as GET plus `aplicado: ["set-prompt", …]` and, after a `prompt` change, `nota`
24
24
  telling whether a `restart` applies (machine) or the identity simply enters on the next
@@ -87,7 +87,7 @@ curl -s "$B/bundle" -H "Authorization: Bearer $GHOSTY_AGENT_TOKEN" > agent.json
87
87
  `{ "connectors": { "google-drive": {} } }` grants the agent access to that account connector; `null` removes it. Drive also takes `files` (ids) and `clientWrite`. Owner only: `403` with a `gat_` token.
88
88
 
89
89
  ## PATCH / — tools per channel, bag
90
- `{ "tools": { "canales": { "wa": { "actualizar_identidad": false } }, "gs": { "crm_buscar": false } } }` turns a tool off in one channel or everywhere (`true` turns it back on). `{ "bag": "MiniGhosty" | null }` = which token bag of its workspace it spends from; goes alone, Ghosty staff only.
90
+ `{ "tools": { "canales": { "wa": { "identity_update": false } }, "gs": { "crm_buscar": false } } }` turns a tool off in one channel or everywhere (`true` turns it back on). `{ "bag": "MiniGhosty" | null }` = which token bag of its workspace it spends from; goes alone, Ghosty staff only.
91
91
 
92
92
  ## PUT /avatar
93
93
  Raw image (`content-type: image/png|jpeg|webp`, ≤ 10 MB) or `{ "url": "https://…" }`. Stored at `img.ghosty.studio`, and written to every Teams where the agent is active. → `{ avatar, workspaces[] }`. Not with a `gat_` token.
@@ -99,7 +99,7 @@ Deletes the agent and its machines. Owner only (`403` for an editor or a `gat_`
99
99
  Owner session only (`403` with a `gat_` token, an editor or a shared admin). `rotate` → `{ agentId, token }`: the new `gat_` token, shown ONCE; the old one stops working. `revoke` → `{ agentId, revoked: true, hadToken }`: the editor channel closes. On ACP engines both restart the relay and cut the turn in flight; `502` = it couldn't reach the box and nothing changed. CLI: `ghosty agents token rotate|revoke`.
100
100
 
101
101
  ## GET|POST|DELETE /memory
102
- Rules its team taught it and its promos (the agent page → Memory). `GET` → `{ notes: [{ id, scope, content, source, createdAt }], promos: [{ id, title, details, imageUrl, validFrom, validUntil, active, live }] }` (`scope` `""` = every channel; `live` = offered today, CDMX). `POST { content, scope?, replaces? }` → `{ note }` (`replaces` edits that rule; without `scope` it keeps its channel). `POST { promo: { id?, title?, details?, validFrom?, validUntil?, active?, imageUrl? } }` → `{ promo, created, warning? }` (`201` when created; with `id` only the given fields change). `DELETE { id }` → `{ deleted, kind: note|promo }`. Works with the `gat_` token too. CLI: `ghosty agents memory`.
102
+ Rules its team taught it and its promos (the agent page → Memory). `GET` → `{ notes: [{ id, scope, content, source, createdAt, delivered }], promos: [{ id, title, details, imageUrl, validFrom, validUntil, active, live }] }` (`scope` `""` = every channel; `delivered: false` = the rule no longer fits in the turn and does not reach the agent: move old rules to the prompt; `live` = offered today, CDMX). `POST { content, scope?, replaces? }` → `{ note }` (`replaces` edits that rule; without `scope` it keeps its channel). `POST { promo: { id?, title?, details?, validFrom?, validUntil?, active?, imageUrl? } }` → `{ promo, created, warning? }` (`201` when created; with `id` only the given fields change). `DELETE { id }` → `{ deleted, kind: note|promo }`. Works with the `gat_` token too. CLI: `ghosty agents memory`.
103
103
 
104
104
  ## POST /transfer
105
105
  Owner (or Ghosty staff) session, never `gat_`. `{ to: email, workspace?: their-slug, dryRun? }` hands the agent over with its box, memory and files → `{ kind: "transfer", to, workspace, workspaces, warnings, transferred? }`. `{ workspace: slug|"none", dryRun?, force? }` moves it between your workspaces → `{ kind: "move", from, to, warnings, steps, failed }`. Always `dryRun: true` first and show it; `400` with a readable reason (no account, plan limit, model over the plan, @handle taken, busy box). CLI: `ghosty agents transfer`.
@@ -124,6 +124,18 @@ closes the run.
124
124
  Exit code 1 from `--wait`/`watch` only means the timeout ran out; run `watch` again. The user can
125
125
  always follow along in the room; `ghosty factory runs card <n>` brings the live card down.
126
126
 
127
+ ## 6. Without a room: send one task to the cloud
128
+
129
+ For a single, well-scoped task you do not need a workspace or a room. Write it as a GitHub issue with checkable acceptance criteria and send it to a work box in your own account:
130
+
131
+ ```bash
132
+ ghosty dev run --issue 12 --watch # inside the clone; or: ghosty dev run OWNER/REPO --title … --goal … --criteria …
133
+ ghosty dev watch <id> # one blocking call; returns when it needs you or closes
134
+ ghosty dev merge <id> # only with green CI and a clean review of the same commit
135
+ ```
136
+
137
+ It needs your GitHub connected to your Ghosty account, and it works on what is pushed to GitHub, not on your local uncommitted changes. Zero checks is not green, and review is capped at 3 rounds.
138
+
127
139
  ## Verify
128
140
 
129
141
  - Every command in `AGENTS.md` runs.