futuresearch · nikosbosse · Feb 18, 2026 · Feb 16, 2026 · Feb 17, 2026 · Feb 18, 2026
diff --git a/docs-site/scripts/check-links.py b/docs-site/scripts/check-links.py
@@ -22,6 +22,12 @@
 # GitHub blob URLs pointing into this repo are checked as local files
 REPO_BLOB_PREFIX = "https://github.com/futuresearch/everyrow-sdk/blob/main/"
 
+# Colab URLs pointing into this repo are checked as local files
+REPO_COLAB_PREFIX = (
+    "https://colab.research.google.com/github/"
+    "futuresearch/everyrow-sdk/blob/main/"
+)
+
 # Git LFS media URLs — the correct way to link to LFS-tracked files.
 # These are checked as local files instead of fetching from GitHub.
 REPO_LFS_PREFIX = (
@@ -186,6 +192,16 @@ def check_file(
                     )
                 continue
 
+            # Colab links to this repo: check the notebook exists locally.
+            if url_without_fragment.startswith(REPO_COLAB_PREFIX):
+                rel_path = url_without_fragment[len(REPO_COLAB_PREFIX) :]
+                if not (REPO_ROOT / rel_path).exists():
+                    errors.append(
+                        f"  {page_label}: file not found for {href!r}"
+                        f" (expected {rel_path})"
+                    )
+                continue
+
             if url_without_fragment in SKIPPED_URLS:
                 continue
 

diff --git a/docs-site/src/app/case-studies/[slug]/page.tsx b/docs-site/src/app/case-studies/[slug]/page.tsx
@@ -1,5 +1,6 @@
 import { notFound } from "next/navigation";
 import { DocsLayout } from "@/components/DocsLayout";
+import { NotebookActions } from "@/components/NotebookActions";
 import { getNavigation } from "@/utils/docs";
 import { getNotebookBySlug, getNotebookSlugs } from "@/utils/notebooks";
 
@@ -50,6 +51,7 @@ export default async function NotebookPage({ params }: PageProps) {
 
   return (
     <DocsLayout navigation={navigation}>
+      <NotebookActions slug={slug} />
       <article
         className="notebook-content"
         dangerouslySetInnerHTML={{ __html: notebook.html }}

diff --git a/docs-site/src/components/NotebookActions.tsx b/docs-site/src/components/NotebookActions.tsx
@@ -0,0 +1,56 @@
+const GITHUB_BASE =
+  "https://github.com/futuresearch/everyrow-sdk/blob/main/docs/case_studies";
+const COLAB_BASE =
+  "https://colab.research.google.com/github/futuresearch/everyrow-sdk/blob/main/docs/case_studies";
+
+interface NotebookActionsProps {
+  slug: string;
+}
+
+export function NotebookActions({ slug }: NotebookActionsProps) {
+  const githubUrl = `${GITHUB_BASE}/${slug}/notebook.ipynb`;
+  const colabUrl = `${COLAB_BASE}/${slug}/notebook.ipynb`;
+
+  return (
+    <div className="notebook-actions">
+      <a
+        href={githubUrl}
+        target="_blank"
+        rel="noopener noreferrer"
+        className="notebook-action-link"
+      >
+        <svg
+          width="16"
+          height="16"
+          viewBox="0 0 24 24"
+          fill="currentColor"
+          aria-hidden="true"
+        >
+          <path d="M12 0C5.37 0 0 5.37 0 12c0 5.31 3.435 9.795 8.205 11.385.6.105.825-.255.825-.57 0-.285-.015-1.23-.015-2.235-3.015.555-3.795-.735-4.035-1.41-.135-.345-.72-1.41-1.23-1.695-.42-.225-1.02-.78-.015-.795.945-.015 1.62.87 1.845 1.23 1.08 1.815 2.805 1.305 3.495.99.105-.78.42-1.305.765-1.605-2.67-.3-5.46-1.335-5.46-5.925 0-1.305.465-2.385 1.23-3.225-.12-.3-.54-1.53.12-3.18 0 0 1.005-.315 3.3 1.23.96-.27 1.98-.405 3-.405s2.04.135 3 .405c2.295-1.56 3.3-1.23 3.3-1.23.66 1.65.24 2.88.12 3.18.765.84 1.23 1.905 1.23 3.225 0 4.605-2.805 5.625-5.475 5.925.435.375.81 1.095.81 2.22 0 1.605-.015 2.895-.015 3.3 0 .315.225.69.825.57A12.02 12.02 0 0 0 24 12c0-6.63-5.37-12-12-12z" />
+        </svg>
+        View source
+      </a>
+      <a
+        href={colabUrl}
+        target="_blank"
+        rel="noopener noreferrer"
+        className="notebook-action-link notebook-action-colab"
+      >
+        <svg
+          width="16"
+          height="16"
+          viewBox="0 0 24 24"
+          fill="none"
+          stroke="currentColor"
+          strokeWidth="2"
+          strokeLinecap="round"
+          strokeLinejoin="round"
+          aria-hidden="true"
+        >
+          <polygon points="5 3 19 12 5 21 5 3" />
+        </svg>
+        Run in Colab
+      </a>
+    </div>
+  );
+}
diff --git a/docs-site/src/styles/notebook.css b/docs-site/src/styles/notebook.css
@@ -1,3 +1,38 @@
+/* Notebook action buttons (View source, Run in Colab) */
+.notebook-actions {
+  display: flex;
+  gap: 0.75rem;
+  margin-bottom: 1rem;
+  justify-content: flex-end;
+}
+
+.notebook-action-link {
+  display: inline-flex;
+  align-items: center;
+  gap: 0.375rem;
+  padding: 0.375rem 0.75rem;
+  font-size: 0.8125rem;
+  font-weight: 500;
+  color: var(--muted);
+  border: 1px solid var(--border);
+  border-radius: 0.375rem;
+  text-decoration: none;
+  transition: color 0.15s, border-color 0.15s, background-color 0.15s;
+}
+
+.notebook-action-link:hover {
+  color: var(--foreground);
+  border-color: var(--foreground);
+  background-color: var(--code-bg);
+  text-decoration: none;
+}
+
+.notebook-action-colab:hover {
+  color: var(--accent);
+  border-color: var(--accent);
+  background-color: var(--accent-light);
+}
+
 /*
  * Notebook styles - scoped to .notebook-content
  * Based on Jupyter's default styling but minimal

diff --git a/docs/case_studies/dedupe-crm-company-records/notebook.ipynb b/docs/case_studies/dedupe-crm-company-records/notebook.ipynb
@@ -3,7 +3,11 @@
   {
    "cell_type": "markdown",
    "metadata": {},
-   "source": "# How to use LLMs to deduplicate CRM Data\n\nThis notebook demonstrates how to use the everyrow SDK's `dedupe` operation to deduplicate messy CRM data using AI-powered semantic matching."
+   "source": [
+    "# How to use LLMs to deduplicate CRM Data\n",
+    "\n",
+    "This notebook demonstrates how to use the everyrow SDK's `dedupe` operation to deduplicate messy CRM data using AI-powered semantic matching."
+   ]
   },
   {
    "cell_type": "markdown",
@@ -19,7 +23,19 @@
    "execution_count": null,
    "metadata": {},
    "outputs": [],
-   "source": "from datetime import datetime\nfrom textwrap import dedent\n\nimport pandas as pd\nfrom dotenv import load_dotenv\n\nfrom everyrow.ops import dedupe\n\nload_dotenv()"
+   "source": [
+    "# !pip install everyrow\n",
+    "from datetime import datetime\n",
+    "from textwrap import dedent\n",
+    "\n",
+    "import pandas as pd\n",
+    "# load API key from environment/.env file or set it directly in the notebook\n",
+    "from dotenv import load_dotenv\n",
+    "load_dotenv()\n",
+    "# import os\n",
+    "# os.environ[\"EVERYROW_API_KEY\"] = \"get an API key on everyrow.io. $20 free\"\n",
+    "from everyrow.ops import dedupe\n"
+   ]
   },
   {
    "cell_type": "markdown",
@@ -188,7 +204,7 @@
     }
    ],
    "source": [
-    "data = pd.read_csv(\"../data/case_01_crm_data.csv\", engine=\"pyarrow\")\n",
+    "data = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/data/case_01_crm_data.csv\", engine=\"pyarrow\")\n",
     "\n",
     "print(f\"Total records: {len(data)}\")\n",
     "data.sort_values(by=\"company_name\").head(15)"
@@ -219,7 +235,18 @@
    "execution_count": null,
    "metadata": {},
    "outputs": [],
-   "source": "equivalence_relation = dedent(\"\"\"\n    Two entries are duplicates if they include data for the same legal entity.\n\"\"\")\n\nprint(\"Deduplicating CRM data...\\n\")\n\nresult = await dedupe(\n    input=data,\n    equivalence_relation=equivalence_relation,\n)"
+   "source": [
+    "equivalence_relation = dedent(\"\"\"\n",
+    "    Two entries are duplicates if they include data for the same legal entity.\n",
+    "\"\"\")\n",
+    "\n",
+    "print(\"Deduplicating CRM data...\\n\")\n",
+    "\n",
+    "result = await dedupe(\n",
+    "    input=data,\n",
+    "    equivalence_relation=equivalence_relation,\n",
+    ")"
+   ]
   },
   {
    "cell_type": "markdown",
@@ -584,6 +611,9 @@
   }
  ],
  "metadata": {
+  "everyrow": {
+   "description": "Python notebook cleaning 500 CRM records with inconsistent company names, missing contacts, and partial email matches. Uses everyrow's dedupe() with a plain-English equivalence relation to find and group semantic duplicates."
+  },
   "kernelspec": {
    "display_name": ".venv",
    "language": "python",
@@ -601,10 +631,7 @@
    "pygments_lexer": "ipython3",
    "version": "3.12.11"
   },
-  "language_version": "3.12",
-  "everyrow": {
-   "description": "Python notebook cleaning 500 CRM records with inconsistent company names, missing contacts, and partial email matches. Uses everyrow's dedupe() with a plain-English equivalence relation to find and group semantic duplicates."
-  }
+  "language_version": "3.12"
  },
  "nbformat": 4,
  "nbformat_minor": 4

diff --git a/docs/case_studies/llm-powered-merging-at-scale/notebook.ipynb b/docs/case_studies/llm-powered-merging-at-scale/notebook.ipynb
@@ -39,6 +39,12 @@
    "metadata": {},
    "outputs": [],
    "source": [
+    "# !pip install everyrow\n",
+    "# load API key from environment/.env file or set it directly in the notebook\n",
+    "from dotenv import load_dotenv\n",
+    "load_dotenv()\n",
+    "# import os\n",
+    "# os.environ[\"EVERYROW_API_KEY\"] = \"get an API key on everyrow.io. $20 free\"\n",
     "import numpy as np\n",
     "import pandas as pd\n",
     "from everyrow.ops import merge\n",
@@ -115,8 +121,8 @@
     }
    ],
    "source": [
-    "left_df = pd.read_csv(\"merge_websites_input_left_2246.csv\")\n",
-    "right_df = pd.read_csv(\"merge_websites_input_right_2246.csv\")\n",
+    "left_df = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-powered-merging-at-scale/merge_websites_input_left_2246.csv\")\n",
+    "right_df = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-powered-merging-at-scale/merge_websites_input_right_2246.csv\")\n",
     "\n",
     "print(f\"Left table: {len(left_df)} rows\")\n",
     "left_df.head(3)"
@@ -240,8 +246,8 @@
     "for n in [100, 200, 400, 800, 1600, 2246]:\n",
     "    result = await merge(\n",
     "        task=\"Match each person to their website(s).\",\n",
-    "        left_table=pd.read_csv(f\"merge_websites_input_left_{n}.csv\"),\n",
-    "        right_table=pd.read_csv(f\"merge_websites_input_right_{n}.csv\"),\n",
+    "        left_table=pd.read_csv(f\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-powered-merging-at-scale/merge_websites_input_left_{n}.csv\"),\n",
+    "        right_table=pd.read_csv(f\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-powered-merging-at-scale/merge_websites_input_right_{n}.csv\"),\n",
     "    )\n",
     "    print(f\"n={n}\")\n",
     "    print(\"num of matched rows:\", len(result.data))\n",
@@ -307,7 +313,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "results_df = pd.read_csv(\"merge_websites_output_800.csv\")"
+    "results_df = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-powered-merging-at-scale/merge_websites_output_800.csv\")"
    ]
   },
   {

diff --git a/docs/case_studies/llm-powered-screening-at-scale/notebook.ipynb b/docs/case_studies/llm-powered-screening-at-scale/notebook.ipynb
@@ -35,15 +35,19 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from dotenv import load_dotenv\n",
+    "# !pip install everyrow\n",
     "import pandas as pd\n",
     "from everyrow import create_session\n",
     "from everyrow.ops import screen\n",
     "\n",
     "pd.set_option(\"display.max_colwidth\", None)\n",
     "\n",
     "\n",
-    "load_dotenv()"
+    "# load API key from environment/.env file or set it directly in the notebook\n",
+    "from dotenv import load_dotenv\n",
+    "load_dotenv()\n",
+    "# import os\n",
+    "# os.environ[\"EVERYROW_API_KEY\"] = \"get an API key on everyrow.io. $20 free\"\n"
    ]
   },
   {
@@ -217,7 +221,7 @@
     }
    ],
    "source": [
-    "fda_product_recalls_df = pd.read_csv(\"fda_product_recalls.csv\")\n",
+    "fda_product_recalls_df = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-powered-screening-at-scale/fda_product_recalls.csv\")\n",
     "\n",
     "# Filter to recalls where center_classification_date is after 2021-08-01 to get a dataset with ≈10k rows\n",
     "fda_product_recalls_df[\"center_classification_date\"] = pd.to_datetime(fda_product_recalls_df[\"center_classification_date\"], errors=\"coerce\")\n",
@@ -311,7 +315,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "results_df = pd.read_csv(\"Screen child product recalls.csv\") # download from https://everyrow.io/sessions/df145a50-2dfd-48c6-97ed-6f82a82bca66"
+    "results_df = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-powered-screening-at-scale/Screen%20child%20product%20recalls.csv\") # download from https://everyrow.io/sessions/df145a50-2dfd-48c6-97ed-6f82a82bca66"
    ]
   },
   {

diff --git a/docs/case_studies/llm-web-research-agents-at-scale/notebook.ipynb b/docs/case_studies/llm-web-research-agents-at-scale/notebook.ipynb
@@ -39,7 +39,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from dotenv import load_dotenv\n",
+    "# !pip install everyrow\n",
     "import pandas as pd\n",
     "from pydantic import BaseModel, Field\n",
     "from everyrow import create_session\n",
@@ -48,7 +48,11 @@
     "pd.set_option(\"display.max_colwidth\", None)\n",
     "\n",
     "\n",
-    "load_dotenv()"
+    "# load API key from environment/.env file or set it directly in the notebook\n",
+    "from dotenv import load_dotenv\n",
+    "load_dotenv()\n",
+    "# import os\n",
+    "# os.environ[\"EVERYROW_API_KEY\"] = \"get an API key on everyrow.io. $20 free\"\n"
    ]
   },
   {
@@ -166,7 +170,7 @@
     }
    ],
    "source": [
-    "input_df = pd.read_csv(\"regulatory_status_results.csv\", usecols=[\"row_id\", \"trade_name\", \"ingredient\", \"applicant\", \"strength\", \"dosage_form\"])\n",
+    "input_df = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-web-research-agents-at-scale/regulatory_status_results.csv\", usecols=[\"row_id\", \"trade_name\", \"ingredient\", \"applicant\", \"strength\", \"dosage_form\"])\n",
     "print(f\"{len(input_df):,} drug products\")\n",
     "print(f\"Columns: {list(input_df.columns)}\")\n",
     "input_df.head(5)"
@@ -356,7 +360,7 @@
     }
    ],
    "source": [
-    "results_df = pd.read_csv(\"regulatory_status_results.csv\")\n",
+    "results_df = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/case_studies/llm-web-research-agents-at-scale/regulatory_status_results.csv\")\n",
     "print(f\"Total rows: {len(results_df):,}\")\n",
     "print(f\"Rows with results: {results_df['regulatory_status'].notna().sum():,}\")\n",
     "print(f\"Failed rows: {results_df['regulatory_status'].isna().sum()}\")\n",
@@ -848,4 +852,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 5
-}
+}
diff --git a/docs/case_studies/match-software-vendors-to-requirements/notebook.ipynb b/docs/case_studies/match-software-vendors-to-requirements/notebook.ipynb
@@ -25,8 +25,12 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!pip install everyrow\n",
-    "%env EVERYROW_API_KEY=your_api_key"
+    "# !pip install everyrow\n",
+    "# load API key from environment/.env file or set it directly in the notebook\n",
+    "from dotenv import load_dotenv\n",
+    "load_dotenv()\n",
+    "# import os\n",
+    "# os.environ[\"EVERYROW_API_KEY\"] = \"get an API key on everyrow.io. $20 free\"\n"
    ]
   },
   {
@@ -168,7 +172,7 @@
     "from everyrow.generated.models import LLMEnum\n",
     "\n",
     "# Load dataset: 438 S&P 500 companies\n",
-    "data = pd.read_csv(\"../data/companies.csv\")\n",
+    "data = pd.read_csv(\"https://media.githubusercontent.com/media/futuresearch/everyrow-sdk/main/docs/data/companies.csv\")\n",
     "print(f\"Dataset: {data.shape[0]} companies, {data.shape[1]} columns\")\n",
     "data.head()"
    ]