fimo 2.14.1 → 2.15.0-staging.1
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/assets/skills/fimo-cli/SKILL.md +1 -0
- package/assets/skills/fimo-cli/references/account.md +1 -1
- package/assets/skills/fimo-cli/references/agents.md +43 -0
- package/assets/skills/fimo-cli/references/deploy.md +1 -1
- package/assets/skills/fimo-studio/SKILL.md +1 -1
- package/assets/skills/fimo-studio/references/ai-chat/agents.md +1 -1
- package/assets/skills/fimo-studio/references/ai-chat/default-agents.md +2 -2
- package/assets/skills/fimo-studio/references/ai-chat.md +1 -1
- package/assets/skills/fimo-studio/references/billing-and-credits.md +1 -1
- package/assets/skills/fimo-studio/references/cms.md +1 -0
- package/assets/skills/fimo-studio/references/preview-and-responsive.md +1 -1
- package/dist/cli/bundle.json +2 -2
- package/dist/cli/index.js +97 -75
- package/package.json +8 -8
- package/release.json +2 -2
- package/templates/astro/package.json +3 -3
- package/templates/nextjs/package.json +4 -4
- package/templates/react-router/package.json +5 -5
- package/templates/sveltekit/package.json +3 -3
- package/templates/vite/package.json +5 -5
|
@@ -70,6 +70,7 @@ Load a reference with `fimo skills get fimo-cli --ref <name>`. "+ fimo-code `<re
|
|
|
70
70
|
- Invite people. After the first deploy, suggest it when other people work on the site → `invite`
|
|
71
71
|
- AI credit balance (read-only) → `credits`
|
|
72
72
|
- Scheduled commands → `schedules`
|
|
73
|
+
- The site's Studio agents and chats: list, read, send a message, wait for the answer, search, run a routine → `agents`
|
|
73
74
|
- Set up an AI agent, install or sync Fimo skills, `fimo rules sync` → `agent-setup`
|
|
74
75
|
- Update Fimo packages → `update`
|
|
75
76
|
- Report a Fimo problem → `support`
|
|
@@ -41,7 +41,7 @@ fimo invite [--target organization|project] [--email <email>] [--role <role>]
|
|
|
41
41
|
## Referral
|
|
42
42
|
|
|
43
43
|
```bash
|
|
44
|
-
fimo referral link # print your referral link
|
|
44
|
+
fimo referral link # print your referral link; REFERRAL_NOT_ELIGIBLE when the account cannot refer yet
|
|
45
45
|
fimo referral stats # show your referral stats
|
|
46
46
|
```
|
|
47
47
|
|
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
# Agents and chats
|
|
2
|
+
|
|
3
|
+
`fimo agents` reads and talks to the site's Studio agents and chats. An agent is a chat with a job and routines; every command below takes an agent or a regular chat, by ID or exact title (pass the ID when two share a title). A message you send is a message from the signed-in person: everyone on the site sees it in Studio.
|
|
4
|
+
|
|
5
|
+
## Common flow
|
|
6
|
+
|
|
7
|
+
```bash
|
|
8
|
+
fimo agents list --all --json # agents and chats: id, kind, title, newest activity first
|
|
9
|
+
fimo agents send "Content" "Draft a post about our pricing" --json # returns once the server accepted it
|
|
10
|
+
fimo agents read <chat-id> --wait --json # waits for the running turn, then prints
|
|
11
|
+
fimo agents send "Content" "Shorter, please" --wait --json # sends, waits, prints the answer and its credits
|
|
12
|
+
fimo agents send --new "Ideas for the blog" --title "Blog ideas" --json # a new regular chat
|
|
13
|
+
fimo agents search "pricing page" --json # message ids, then:
|
|
14
|
+
fimo agents read <chat-id> --around <message-id> --json
|
|
15
|
+
fimo agents run "SEO" "Weekly checkup" --wait --json # starts a routine now
|
|
16
|
+
```
|
|
17
|
+
|
|
18
|
+
- **`send` returns at once.** A turn can take many minutes, so `send` returns when the server has the message, with `data.chat.id`, `data.messageId`, `data.state` (`started`, `queued` behind a running turn, or `already_sent`) and `data.next`, the command that reads the answer. Returning never stops the turn. Run `data.next` (`read <id> --wait`) to get the answer. Use `send --wait` only when your shell allows a long command.
|
|
19
|
+
- **Message input:** the second argument, or `--file <path>`, or `--stdin`. `--model <key>` picks the model for that turn; Studio's default is `auto`.
|
|
20
|
+
- **`read`** prints the newest messages oldest first (`--limit`, default 20; `data.nextCursor` with `--cursor` for older ones; `--around <message-id>` for the messages around a search hit). Each agent message shows its answer: the text after the agent marked where its reply starts, and one line per tool call. `--full` also prints the work before it. `data.running` says whether a turn runs now, `data.queue` what waits, and `data.openQuestions` what the agent asked a person. Do not answer those questions yourself: tell the person to answer them in Studio.
|
|
21
|
+
- **Credits:** every command puts the Fimo credits it used in `data.credits` and ends its text output with a `Credits:` line. Each agent message carries `credits`, the credits its turn used, and `read` totals the answers it shows. `send` and `run` without `--wait` give `null`: the turn is not counted until it ends. `search` gives `0`. `runs` shows each run's credits. `null` always means no settled bill yet.
|
|
22
|
+
- **`run <agent> <routine>`** starts a routine and returns at once (`data.started`; `--message <text>` gives the run a message). `--wait` waits for the run this call started and prints `data.run`: status, outcome, summary, duration and credits. Follow a run later with `fimo agents runs <agent> <routine> --limit 1`.
|
|
23
|
+
- **`search <words>`** is full-text over what the site's chats show people: `--chat <agent-or-chat>`, `--after <ISO time>`, `--before <ISO time>`, `--limit <1-50>`. Each hit has `chatId`, `messageId`, `authorKind`, `snippet` and `read`, the command that reads around it.
|
|
24
|
+
- **Ctrl-C** during a wait stops only the waiting. The turn goes on; the error's hint is the command that reads it later.
|
|
25
|
+
|
|
26
|
+
## Refusals
|
|
27
|
+
|
|
28
|
+
| `error.code` | What happened → next step |
|
|
29
|
+
| ------------------------ | ------------------------------------------------------------------------------------------ |
|
|
30
|
+
| `CHAT_NOT_FOUND` | No agent or chat with that ID or title on this site → `fimo agents list --all` |
|
|
31
|
+
| `AMBIGUOUS_CHAT` | Two chats share the title → use the ID |
|
|
32
|
+
| `CHAT_TAKES_NO_MESSAGES` | A routine's or task's thread is read-only → send to its agent |
|
|
33
|
+
| `INSUFFICIENT_CREDITS` | The message was kept, but no credits are left for its turn → add credits in Studio |
|
|
34
|
+
| `RATE_LIMITED` | 20 messages in a minute → wait a minute |
|
|
35
|
+
| `EMPTY_MESSAGE` | Nothing to send |
|
|
36
|
+
| `AGENT_NOT_LIVE` | The agent is still being set up → finish it in Studio |
|
|
37
|
+
| `NOT_ON_PLAN` | The routine needs Pro |
|
|
38
|
+
| `OUT_OF_AGENT_HOURS` | No agent-hours left until the reset in the message |
|
|
39
|
+
| `API_UNSUPPORTED` | The Fimo API is older than this CLI and has no chat search |
|
|
40
|
+
| `WAIT_INTERRUPTED` | You stopped waiting; the turn goes on → the hint's command |
|
|
41
|
+
| `PERMISSION_DENIED` | Your role cannot send in this site's chats (a project token from `fimo new` cannot either) |
|
|
42
|
+
|
|
43
|
+
Fimo's own agents read with `list`, `show`, `runs`, `read` and `search` from their sandbox; `send` and `run` refuse them (`capability denied: agents:prompt`), because a message from the CLI speaks for a person. The hosted MCP has none of these commands.
|
|
@@ -22,7 +22,7 @@ fimo preview url --json # cu
|
|
|
22
22
|
- **Publish only when the user says to publish, ship, or go live.** Publishing makes the site public.
|
|
23
23
|
- Local changes to go live: `fimo deploy --publish`. The revision already on Fimo is the release: `fimo publish` (it never pushes).
|
|
24
24
|
- Check the reported source commit and warnings before you report success. If local work is not in the build, or the built commit is unknown, follow the hint.
|
|
25
|
-
- A failed publication usually gives its id (`error.details.publication_id`; else find it with `fimo publish list`). Run `fimo publish inspect <id> --json` for the status and cause, then `fimo publish logs <id> --json` for the build log. The hosting provider can delete old logs: then report that, and do not publish again only to get logs. Older attempts: `fimo publish list --cursor <cursor>` or `--all-pages`.
|
|
25
|
+
- A failed publication usually gives its id (`error.details.publication_id`; else find it with `fimo publish list`). Run `fimo publish inspect <id> --json` for the status and cause, then `fimo publish logs <id> --json` for the build log (a Studio chat agent cannot run it; it reads the cause from `publish inspect`). The hosting provider can delete old logs: then report that, and do not publish again only to get logs. Older attempts: `fimo publish list --cursor <cursor>` or `--all-pages`.
|
|
26
26
|
|
|
27
27
|
## A failed preview deploy
|
|
28
28
|
|
|
@@ -35,7 +35,7 @@ Use plain, friendly language. Describe what users see, where they click, and wha
|
|
|
35
35
|
- **Header (desktop):** the site menu (site name; History is here) and the Branch switcher, then the workspaces: Chats, Getting started, Preview, Content, Assets, Analytics, Settings. Only workspaces the user can access appear. Workspaces that do not fit move into **More**. On the right: Earn free credits, Invite, Clone, Publish, and Merge on a Branch.
|
|
36
36
|
- **Phone:** bottom tabs for Chat, Preview and Settings. Invite, Earn free credits and Clone ("Work on this site locally") move into the site menu.
|
|
37
37
|
- **Settings:** General, Domains, Knowledge, SEO, Integrations and Languages are listed directly. Models, Environments, Variables, Secrets, Schedules and Webhooks are under **Advanced settings**.
|
|
38
|
-
- **Shortcuts:** Alt+1 Preview, Alt+2 Content, Alt+3 Assets, Alt+5 Settings, Alt+E edit mode, Cmd/Ctrl+K command palette.
|
|
38
|
+
- **Shortcuts:** Alt+1 Preview, Alt+2 Content, Alt+3 Assets, Alt+5 Settings, Alt+E edit mode, Cmd/Ctrl+K command palette, / page list in Preview.
|
|
39
39
|
|
|
40
40
|
When a reference says "open X", X can be inside **More** on a narrow window.
|
|
41
41
|
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
# Agents
|
|
2
2
|
|
|
3
|
-
An agent is a chat with a job: a name, a one-line summary for people, and instructions that Fimo reads at the start of every reply in its chat and every run of its routines. A live agent is shared like any chat and listed under **Agents** in the chat list. What each default routine does is in `ai-chat/default-agents`; triggers, runs, reports and agent-hours are in `ai-chat/routines`.
|
|
3
|
+
An agent is a chat with a job: a name, a one-line summary for people, and instructions that Fimo reads at the start of every reply in its chat and every run of its routines. A live agent is shared like any chat and listed under **Agents** in the chat list. From a terminal, a person or their coding agent lists, reads, messages and searches agents and chats, and runs routines, with `fimo agents` (`fimo-cli` reference `agents`); Fimo's own agents only read there. What each default routine does is in `ai-chat/default-agents`; triggers, runs, reports and agent-hours are in `ai-chat/routines`.
|
|
4
4
|
|
|
5
5
|
## What each plan gets
|
|
6
6
|
|
|
@@ -21,10 +21,10 @@ Alone, it points a link at a new address only after it checked that the address
|
|
|
21
21
|
|
|
22
22
|
- **AI readiness:** llms.txt, the text each page serves before JavaScript, landmarks and heading levels.
|
|
23
23
|
- **Structured data:** valid JSON-LD that fits each page: Organization and WebSite, Article, BreadcrumbList; Product and FAQ where they fit.
|
|
24
|
-
- **AI citation optimization:** a direct answer under each content page's H1,
|
|
24
|
+
- **AI citation optimization:** a direct answer under each content page's H1, question headings on long pages, and proposed new words for a line AI answers hold against it.
|
|
25
25
|
- **AI visibility** (once a week): how often ChatGPT, Gemini, Perplexity and Google AI Mode name and cite the site for about 20 customer questions. It changes nothing.
|
|
26
26
|
|
|
27
|
-
Alone, it writes a missing llms.txt and keeps its lines in step with the pages
|
|
27
|
+
Alone, it writes a missing llms.txt and keeps its lines in step with the pages, their titles and descriptions, adds Organization, WebSite, Article and BreadcrumbList from facts the site already shows (never an invented logo, address, author or rating), and fixes landmarks and heading levels without changing a word. It asks before an opening answer, new headings, FAQ or Product data, or a change in how pages render (unless one setting serves the same pages as HTML and nothing else changes). It never changes who robots.txt lets crawl, AI crawlers included, or invents a claim; asked, it names the crawlers robots.txt keeps out.
|
|
28
28
|
|
|
29
29
|
## Content (6 routines)
|
|
30
30
|
|
|
@@ -52,7 +52,7 @@ Hover over a message. Your messages have **Copy** and **Edit** (change and resen
|
|
|
52
52
|
## Proposals, reviews and merges
|
|
53
53
|
|
|
54
54
|
- Ask chat to keep alternatives as named proposals. Open the Branch picker and choose one under **Proposals** to preview it; each has its own changes. Internal work stays hidden there, even when saved or kept longer. Exploring or merging one keeps the others unless you ask to discard them. Delete a proposal from the Branch picker. Pull is not available for a proposal: ask chat to update it. **New Branch** starts from the proposal's upstream; merge the proposal first to branch from its changes.
|
|
55
|
-
- A request is made at once. When a decision is still yours, such as ideas you asked to pick from or a change a routine found, chat proposes instead: a block with one line per proposal (spinner: still being written, arrow: ready, check: approved, crossed out: discarded). A line opens the proposal beside the chat, or as its own page on a phone: new or changed content as a page of text you can edit (your edits stay in the proposal, nothing is live), and a change to how a page looks or works as that page's preview. **Approve** sends `Approve "<title>"` to the chat, and the agent writes it to your preview, or first makes a change you asked for there; if someone changed the same field since, nothing is overwritten and the agent says what changed. **Discard** removes it at once.
|
|
55
|
+
- A request is made at once. When a decision is still yours, such as ideas you asked to pick from or a change a routine found, chat proposes instead: a block with one line per proposal, its title over the first sentence of why (spinner: still being written, arrow: ready, check: approved, crossed out: discarded). A line opens the proposal beside the chat, or as its own page on a phone: new or changed content as a page of text you can edit (your edits stay in the proposal, nothing is live), and a change to how a page looks or works as that page's preview. On a content proposal, a bar under the header gives the agent's reason and opens to all of it; on a page that exists, the header counts the changed fields, each changed text field marks its new words in green with a **Before** line under it, and a repeated item with a change is open and marked **Changed**. **Approve** sends `Approve "<title>"` to the chat, and the agent writes it to your preview, or first makes a change you asked for there; if someone changed the same field since, nothing is overwritten and the agent says what changed. **Discard** removes it at once.
|
|
56
56
|
- A session proposal saves the session's current work and makes that same session visible and durable. It shows in a strip above the message box (**View changes**, **Open preview**), and the conversation keeps a line with **View changes** where it was proposed. Opening it approves nothing. The Preview and Changes links, also in Slack or an agent report, open the Branch in Studio and need site access.
|
|
57
57
|
- Chat prepares changes separately and merges them into the environment you view; most merge automatically. If the site changed in the same place, a banner above the message box (**Your last changes need attention**) offers **Auto fix** (chat continues in the retained proposal, resolves the conflict and retries the merge; keep chatting has the same effect) and **Review manually**. Discarding leaves the environment unchanged and removes the proposal for good.
|
|
58
58
|
- **Merging does not publish the site.**
|
|
@@ -25,7 +25,7 @@ Open **Earn free credits** from the site header (or the site menu on mobile), th
|
|
|
25
25
|
- **Published** counts people who published their first site, not the number of sites they publish.
|
|
26
26
|
- **AI credits earned** shows inviter rewards already credited to you. It excludes your own bonus for joining through someone else's link; a publication can appear before its reward is credited.
|
|
27
27
|
|
|
28
|
-
Each signup uses one of the five referral slots. The reward unlocks when that person publishes a site at least a day after signing up, and is credited to the referrer's personal organization; publishing does not reopen a slot. Someone who joined through a referral can't refer others,
|
|
28
|
+
Each signup uses one of the five referral slots. The reward unlocks when that person publishes a site at least a day after signing up, and is credited to the referrer's personal organization; publishing does not reopen a slot. Someone who joined through a referral can't refer others, and an account can refer only once it is three days old; in both cases the dialog shows a notice instead of a link, with the date the link starts earning for a new account. The dialog shows signups used and how many people have not published yet. **Terms and Conditions** opens the program terms; **Close** dismisses the dialog.
|
|
29
29
|
|
|
30
30
|
## Buying More Credits
|
|
31
31
|
|
|
@@ -15,6 +15,7 @@ Pages is still a pre-release Studio surface. It is visible in development and st
|
|
|
15
15
|
1. Open **Content** and choose **Pages**, **Collections**, **Singles**, **Forms**, or **Labels & Text** in the left sidebar. Hover over or focus the info icon beside any item for a short explanation.
|
|
16
16
|
2. Pages shows the site's route tree. Select a page to edit its declared sources together; for a dynamic route, choose the concrete entry from the header selector.
|
|
17
17
|
3. Collections and Singles show cards for their schemas. Collection cards show their entry count; Single cards are labeled **Single**.
|
|
18
|
+
4. On Collections, Singles, and Forms, the Cards and List icons at the top right switch between cards and compact rows. Studio remembers the choice for each page and each person.
|
|
18
19
|
|
|
19
20
|
## Editing by Page
|
|
20
21
|
|
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
Open **Preview** (the cursor icon) or press **Alt+1** to see a live version of your site.
|
|
4
4
|
|
|
5
|
-
- For Fimo-integrated sites,
|
|
5
|
+
- For Fimo-integrated sites, select the page name at the top, or press **/**, to open the page list. Type to search by name or URL. Pages are grouped, **Recent** lists the last pages previewed, and a page marked **Dynamic** asks for an entry or a value after it opens. The selected page is in the Studio address, so a reload or a shared link opens the same page and Branch.
|
|
6
6
|
- Click the **refresh** button to reload the preview.
|
|
7
7
|
- Click the **external link** icon to open your site in a new browser tab for a full-screen view.
|
|
8
8
|
|
package/dist/cli/bundle.json
CHANGED