@koda-sl/baker-cli 0.192.2 → 0.195.0-dev.6d2f498f5

package/README.md

@@ -1391,13 +1391,62 @@ baker research web "What is the pricing of monday.com?" --output md
 
 ---
 
+### `baker research fetch "https://url"`
+
+Read one named page that an ordinary web fetch could not. Bot walls and JavaScript-rendered pages are resolved by Firecrawl on your behalf, so there is no escalation ladder to hand-roll: no retry loop, no waiting, no browser.
+
+```bash
+baker research fetch "https://competitor.com/pricing"
+baker research fetch "https://competitor.com/pricing" --full
+baker research fetch "https://app.competitor.com/blog/post" --wait-for 5000
+baker research fetch "https://competitor.com" --format html --whole-page
+```
+
+**Response:**
+
+```json
+{
+  "ok": true,
+  "data": {
+    "url": "https://competitor.com/pricing",
+    "final_url": "https://www.competitor.com/pricing",
+    "title": "Pricing — Competitor",
+    "content_chars": 48213,
+    "content_shown": 20000,
+    "truncated": false,
+    "provider": "firecrawl",
+    "content": "# Pricing\n\n## Starter\n..."
+  },
+  "hints": ["Showing the first 20,000 of 48,213 characters. Re-run the same command with `--full` ..."]
+}
+```
+
+**Flags:**
+
+| Flag           | Description                                                              |
+|----------------|--------------------------------------------------------------------------|
+| `--format`     | `markdown` (default) or `html`                                            |
+| `--wait-for`   | Extra ms (0–30000) to let late content settle                             |
+| `--whole-page` | Include nav, footer and sidebars (default: main content only)             |
+| `--full`       | Return the whole page instead of the first 20,000 characters              |
+
+- `truncated: true` means the page is longer than one read returns, so its tail is missing.
+- Only Firecrawl is contacted — the target host is never re-requested directly, so a site that is rate-limiting or walling you is not hit again.
+- Each URL is cached server-side for 24 hours, so a follow-up `--full` call costs no extra credit.
+- Failures carry `error.fix` with `action: "continue_without"` for pages that will not open — finish the rest of the job and report the gap rather than retrying.
+
+---
+
 ### `baker research advertisers "keyword"`
 
-Find domains competing for a keyword in Google SERPs.
+Returns a ranked list of **domains** competing for one keyword in Google, with average SERP position, relevance rating, estimated traffic value and visibility — domain economics, **NOT ad copy**. Despite the name it returns no headlines, descriptions or creative of any kind; for a competitor's actual ad copy use `baker winning-ads content <adId>` (Meta/LinkedIn only — Google SERP ad copy is not available anywhere in Baker).
+
+Defaults to `--location us` and `--language en` when omitted; the response carries a `query_context` showing which pair was used, so always set both for non-US or non-English markets.
 
 ```bash
 baker research advertisers "running shoes"
 baker research advertisers "crm software" --location uk --limit 10
+baker research advertisers "zapatos" --location es --language spanish
 ```
 
 **Response:**
@@ -1744,6 +1793,7 @@ Research data is cached server-side (shared across all callers). No local cache
 | keyword-gap       | 6 hours | Same                                |
 | relevant-pages    | 6 hours | Traffic data updates weekly         |
 | lighthouse        | 24 hours| Page performance stable day-to-day  |
+| fetch             | 24 hours| Page content rarely changes within a day |
 
 ---
 
@@ -1775,6 +1825,7 @@ Each external source is its own subcommand. Pick the verb that matches the sourc
 | `baker images upscale <imageId>` | Real-ESRGAN super-resolution via backend ($0.05/image, cost-tracked) | n/a (operates on library image) |
 | `baker images crop <file>` | Coordinate-based rectangular extract — local file or URL | n/a |
 | `baker images dimensions <file\|url>` | Read width / height / aspect / format without decoding | n/a |
+| `baker images download <targets>` | Remote URLs and/or library ids → local files, so the local transforms can read them | n/a (writes to disk, no upload) |
 | `baker images tags` | List available image tag names (defaults + company custom tags) | n/a |
 
 **Auto-ingest** runs the full `processImage` pipeline (Gemini describe + Voyage multimodal embed + OpenRouter text embed) on every hit. Override with `--auto-ingest N` (turn on) or `--no-auto-ingest` (turn off where default is on). When auto-ingest succeeds, the matching returned hit uses the Baker-owned URL and keeps the original provider URL as `sourceUrl`. After auto-ingest the next `baker images library` query for the same concept hits the local row.
@@ -2314,6 +2365,30 @@ Response:
 }
 ```
 
+### `baker images download <targets...> [--out <dir-or-file>]`
+
+Remote image URLs and/or library image ids → local files. The missing first half of `source → download → normalize → place`: `normalize` refuses URLs and takes no library ids, so without this the only route to disk was `curl`. Local-only, like `normalize` / `crop` / `dimensions` — bytes go straight to disk, never through the backend.
+
+Targets are space- or comma-separated and may mix both kinds. `--out` is a directory (which must exist) or, with a single target, an exact file path; it defaults to the working directory. Names that would collide are disambiguated, and the extension is derived from the response content type.
+
+```bash
+baker images download https://media.withbaker.com/…/logo.webp
+baker images download j57abc123 j57def456 --out src/pages/pricing/_images/
+baker images download https://…/hero.png --out ./hero.png
+```
+
+Response lists what landed and what didn't — a partial failure is still `ok: true`, with the failures in `failed[]` and a hint saying so. Every target failing is `DOWNLOAD_FAILED`. Passing a path that is already on disk is a `VALIDATION_ERROR` whose `fix` points at using the local transforms directly.
+
+```json
+{
+  "ok": true,
+  "data": {
+    "downloaded": [{ "input": "j57abc123", "output": "hero.png", "bytes": 84213, "contentType": "image/png" }],
+    "failed": []
+  }
+}
+```
+
 ### `baker videos search <query>`
 
 Semantic search videos.
@@ -2709,12 +2784,15 @@ baker actions unlink --blocker <id> --blocked <id>
 
 # Review and edit the ops staged in THIS chat before publish
 baker actions draft                        # show every staged op + footgun warnings
+baker actions draft --output md             # one row per staged op (json default | md | files)
+baker actions draft --output md --fields kind,name,ref,tags   # pick the columns
+baker actions draft --output md --full      # add each op's payload (description, note, reason, blocker)
 baker actions draft remove temp_hero        # drop a staged create (cascades its complete/link ops)
 baker actions draft remove <id> --op complete   # drop a staged op on a real action (--op update|complete|discard)
 baker actions draft clear                   # drop everything staged in this chat
 ```
 
-`baker actions draft` is the detailed, op-by-op view of staged (pre-publish) ops plus footgun `warnings`, and the place to edit the draft (`draft remove`/`clear`). The bucketed `list` and a chat-scoped `status` also surface this chat's draft (staged creates, `draftStatus` markers, `draft` ref status), so `draft` is mainly for the full changelog and edits. Do not stage `complete` to cancel an unwanted action — that publishes it as an already-completed item; remove the staged create (or `discard` a published action) instead.
+`baker actions draft` is the detailed, op-by-op view of staged (pre-publish) ops plus footgun `warnings`, and the place to edit the draft (`draft remove`/`clear`). It takes the same `--output json|md|files` / `--fields` / `--full` trio as the other list commands: `json` (the default) keeps the full envelope with status, count and warnings, while `md` and `files` render one row per op — `kind`, `name`, `ref`, `tags` by default, where `ref` is the handle the next command needs (the tempId for a staged create, the real action id for an op against a published one). An unrecognized `--output` is rejected rather than silently printing nothing. The bucketed `list` and a chat-scoped `status` also surface this chat's draft (staged creates, `draftStatus` markers, `draft` ref status), so `draft` is mainly for the full changelog and edits. Do not stage `complete` to cancel an unwanted action — that publishes it as an already-completed item; remove the staged create (or `discard` a published action) instead.
 
 `baker actions status <ref...>` resolves refs in one request to `/api/actions/status`. It works without `BAKER_CHAT_ID`, but when the CLI has a chat id it includes it so a temp still staged in **that** chat resolves to `status: "draft"` instead of `not_found` (only the caller's own chat draft is consulted). It preserves the backend JSON envelope:
 
@@ -5146,6 +5224,7 @@ baker landing inspiration search "pricing with a monthly/annual toggle" --scope
 baker landing inspiration view <section id>     # full DNA + screenshots + motion filmstrip
 baker landing inspiration code <section id>     # the standalone bundle
 baker landing inspiration page <source id>      # a whole page as a section sequence
+baker landing inspiration sequences "b2b saas pricing page" --scope all   # what follows what, across many pages
 baker landing inspiration add https://linear.app --note "client likes this density"
 baker landing inspiration favorites             # what this company has saved
 baker landing inspiration scrape <url> --out <dir>   # capture any page now, synchronously
@@ -5159,6 +5238,7 @@ baker landing inspiration scrape <url> --out <dir>   # capture any page now, syn
 - **`add` refuses a page Baker already serves**, along with preview and private addresses. For our own pages the source is in the workspace, so a capture would file a screenshot and reconstructed markup next to the real thing — and because the corpus is shared, a live client page admitted here would be readable by every other company.
 - **`add` and `favorite` are recorded on the chat as a "Reference page" change, marked already applied.** The save and the study both happen immediately; there is nothing left for publish to apply and nothing a discard takes back.
 - **Compact by default, `--full` when you have chosen something.** Every list command returns what you need to *pick* a row; the 600-character "why it works" paragraph, the classification facets and the markup ride behind `--full` on `search`, `view`, `page`, `favorites` and `code`. A default `favorites` used to cost more than a default search for a question — "what does this client keep saving?" — that the facets alone answer.
+- **`sequences` answers the ordering question search structurally can't.** Search ranks individual sections and caps each site at three, so a page's top-to-bottom order never survives its results, and `page` recovers exactly one page's order. `sequences` counts adjacency across many pages at once and returns `openers`, `transitions` (`{ from, to, count }`, most frequent first) and `closers`, filtered by `--scope --domain --section-type --limit`. It returns page **ids only** — `page` stays the one way to read a single page — and always reports `pages_considered` alongside `pages_returned`, because a frequency over an unstated denominator is not evidence. `--full` counts over every matching page and prints the whole table.
 - **`page_id` on every search row** is what `page <id>` takes, so "how does this page sequence its sections?" is reachable from a result rather than only right after `add`. `favorites --limit` is capped at 100.
 - **`add` studies in the background; `scrape` returns with the page on disk.** `add` grows the shared library and takes minutes, so it can never answer "build our page like this one" within the same turn. `scrape` runs the identical capture locally and blocks until it finishes, writing section screenshots, standalone markup, a whole-page reproduction and `report.html`. `--no-motion` is the single biggest lever on runtime; `--no-mobile`, `--no-code` and `--no-report` skip further passes.
 - **Filming is the expensive pass, and it runs three sections at a time.** Every moving section is filmed in its own fresh page load — the only way to catch an entrance animation before it fires — so a page with ten moving sections pays for ten full loads. Measured on one heavy page, filming was 74% of the capture. The takes are independent, so they overlap; a page that cannot be captured inside the library's budget now says so instead of timing out silently.

package/dist/{chunk-ISZWNERZ.js → chunk-IDRQDUBA.js}

@@ -7,9 +7,9 @@ import {
   __toESM
 } from "./chunk-YL3HDEIJ.js";
 
-// ../../node_modules/.pnpm/safe-stable-stringify@2.5.0/node_modules/safe-stable-stringify/index.js
+// ../../.pnpm-store/v10/links/@/safe-stable-stringify/2.5.0/810146e81bae4e3a061fe487864f2fde80c4b03b886877dc0f1fffbc6480b67e/node_modules/safe-stable-stringify/index.js
 var require_safe_stable_stringify = __commonJS({
-  "../../node_modules/.pnpm/safe-stable-stringify@2.5.0/node_modules/safe-stable-stringify/index.js"(exports, module) {
+  "../../.pnpm-store/v10/links/@/safe-stable-stringify/2.5.0/810146e81bae4e3a061fe487864f2fde80c4b03b886877dc0f1fffbc6480b67e/node_modules/safe-stable-stringify/index.js"(exports, module) {
     "use strict";
     var { hasOwnProperty } = Object.prototype;
     var stringify = configure2();
@@ -1083,7 +1083,7 @@ function resolveAdaptFormats(params) {
   return params.formats ?? [];
 }
 
-// ../../node_modules/.pnpm/safe-stable-stringify@2.5.0/node_modules/safe-stable-stringify/esm/wrapper.js
+// ../../.pnpm-store/v10/links/@/safe-stable-stringify/2.5.0/810146e81bae4e3a061fe487864f2fde80c4b03b886877dc0f1fffbc6480b67e/node_modules/safe-stable-stringify/esm/wrapper.js
 var import__ = __toESM(require_safe_stable_stringify(), 1);
 var configure = import__.default.configure;
 var wrapper_default = import__.default;
@@ -8376,4 +8376,4 @@ export {
   defaultRegistry,
   createEngineFromEnv
 };
-//# sourceMappingURL=chunk-ISZWNERZ.js.map
+//# sourceMappingURL=chunk-IDRQDUBA.js.map