diff --git a/.github/workflows/live.yml b/.github/workflows/live.yml
index 45f69b5..f12b34e 100644
--- a/.github/workflows/live.yml
+++ b/.github/workflows/live.yml
@@ -70,7 +70,7 @@ jobs:
       - name: Install the package and the Bright Data CLI
         run: |
           pip install -e .
-          npm i -g @brightdata/cli@0.3.4  # pinned: this job holds the API secret
+          npm i -g @brightdata/cli@0.4.0  # pinned: this job holds the API secret
 
       - name: Do the README's links still resolve?
         run: |
@@ -310,7 +310,7 @@ jobs:
       - name: Install the package and, for CLI blocks, the Bright Data CLI
         run: |
           pip install -e .
-          if [ "${{ matrix.lang }}" = "bash" ]; then npm i -g @brightdata/cli@0.3.4; fi
+          if [ "${{ matrix.lang }}" = "bash" ]; then npm i -g @brightdata/cli@0.4.0; fi
       - name: Run block ${{ matrix.n }} exactly as the README shows it
         env:
           N: ${{ matrix.n }}
diff --git a/AGENTS.md b/AGENTS.md
index 9fc4716..36d4965 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -99,9 +99,12 @@ and the result labelled with the dead slug held Reid Hoffman's profile.
 
 ### The Bright Data CLI
 
-The LinkedIn pipelines are `linkedin_person_profile`, `linkedin_company_profile`,
-`linkedin_job_listings`, `linkedin_posts` and `linkedin_people_search`. There is
-no `linkedin_profiles`. Run `bdata pipelines list` rather than guessing a name.
+CLI 0.4.0 (12 Oct 2026) removed pipeline type names such as
+`linkedin_person_profile`: `bdata pipelines <type>` now exits 1. Run a scraper
+by dataset ID and method, `bdata pipelines run gd_l1viktl72bvl7bjuj0
+collect_by_url <url>`. Run `bdata pipelines search linkedin` rather than
+guessing an ID. The README block uses unpinned `npx`, so it always gets the
+newest CLI, whatever `live.yml` pins.
 
 ## The API stalls in waves
 
diff --git a/README.md b/README.md
index f46e639..fefcb2b 100644
--- a/README.md
+++ b/README.md
@@ -385,13 +385,15 @@ once, or use `bdata login --device` over SSH and in CI:
 
 ```bash
 npx -p @brightdata/cli bdata login
-npx -p @brightdata/cli bdata pipelines linkedin_person_profile "https://www.linkedin.com/in/satyanadella/"
+npx -p @brightdata/cli bdata pipelines run gd_l1viktl72bvl7bjuj0 collect_by_url "https://www.linkedin.com/in/satyanadella/"
 ```
 
-`bdata pipelines list` prints every type. The LinkedIn ones are
-`linkedin_person_profile`, `linkedin_company_profile`, `linkedin_job_listings`,
-`linkedin_posts` and `linkedin_people_search`. Each takes URLs, prints JSON,
-and costs one credit per record.
+`bdata pipelines search linkedin` finds every scraper and its ID. The LinkedIn
+ones are people profiles `gd_l1viktl72bvl7bjuj0`, company information
+`gd_l1vikfnt1wgvvqz95w`, job listings `gd_lpfll7v5hcqtkxl6l`, posts
+`gd_lyy3tktm25m4avu764` and people search `gd_m8d03he47z8nwb5xc`. Each takes
+URLs with `collect_by_url`, prints JSON, and costs one credit per record.
+`bdata pipelines info <id>` lists a scraper's other methods.
 
 `npx skills add brightdata/skills` teaches Claude Code, Cursor and Codex these
 commands and the docs, so plain language works afterwards. Full guide: