diff --git a/.nojekyll b/.nojekyll
index 7c3e4f458..b7e6ef93b 100644
--- a/.nojekyll
+++ b/.nojekyll
@@ -1 +1 @@
-9424e06d
\ No newline at end of file
+1c47011a
\ No newline at end of file
diff --git a/eval-logs.html b/eval-logs.html
index 93ecae71a..64df9e5c8 100644
--- a/eval-logs.html
+++ b/eval-logs.html
@@ -1097,7 +1097,7 @@ <h3 class="anchored" data-anchor-id="reading-logs">Reading Logs</h3>
     </div>
   </div>
 </footer>
-<script>var lightboxQuarto = GLightbox({"selector":".lightbox","loop":false,"closeEffect":"zoom","descPosition":"bottom","openEffect":"zoom"});
+<script>var lightboxQuarto = GLightbox({"loop":false,"descPosition":"bottom","openEffect":"zoom","selector":".lightbox","closeEffect":"zoom"});
 (function() {
   let previousOnload = window.onload;
   window.onload = () => {
diff --git a/examples.html b/examples.html
index f2b7949c6..ba0708ccb 100644
--- a/examples.html
+++ b/examples.html
@@ -444,7 +444,7 @@ <h2 class="anchored" data-anchor-id="sec-security-guide">Security Guide</h2>
 <section id="setup" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="setup">Setup</h3>
 <p>We’ll start by importing the functions we need from Inspect and defining a system message that orients the model to its role as a computer security expert.</p>
-<div id="a866fd47" class="cell">
+<div id="02b9580c" class="cell">
 <div class="sourceCode cell-code" id="cb1"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb1-1"><a href="#cb1-1" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai <span class="im">import</span> Task, <span class="bu">eval</span>, task</span>
 <span id="cb1-2"><a href="#cb1-2" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.dataset <span class="im">import</span> example_dataset</span>
 <span id="cb1-3"><a href="#cb1-3" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.scorer <span class="im">import</span> model_graded_fact</span>
@@ -461,7 +461,7 @@ <h3 class="unlisted anchored" data-anchor-id="setup">Setup</h3>
 <section id="eval" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="eval">Eval</h3>
 <p>Discerning whether the correct security guidance was provided by the model might provide difficult using only text matching algorithms. Here we use a model to read the response and assess the quality of the answer.</p>
-<div id="bf21bfbb" class="cell">
+<div id="4b70a6f8" class="cell">
 <div class="sourceCode cell-code" id="cb2"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb2-1"><a href="#cb2-1" aria-hidden="true" tabindex="-1"></a><span class="at">@task</span></span>
 <span id="cb2-2"><a href="#cb2-2" aria-hidden="true" tabindex="-1"></a><span class="kw">def</span> security_guide():</span>
 <span id="cb2-3"><a href="#cb2-3" aria-hidden="true" tabindex="-1"></a>    <span class="cf">return</span> Task(</span>
@@ -491,7 +491,7 @@ <h2 class="anchored" data-anchor-id="sec-hellaswag">HellaSwag</h2>
 <section id="setup-1" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="setup-1">Setup</h3>
 <p>We’ll start by importing the functions we need from Inspect, defining a system message, and writing a function to convert dataset records to samples (we need to do this to convert the index-based label in the dataset to a letter).</p>
-<div id="79993a76" class="cell">
+<div id="1f27d09d" class="cell">
 <div class="sourceCode cell-code" id="cb4"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb4-1"><a href="#cb4-1" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai <span class="im">import</span> Task, <span class="bu">eval</span>, task</span>
 <span id="cb4-2"><a href="#cb4-2" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.dataset <span class="im">import</span> Sample, hf_dataset</span>
 <span id="cb4-3"><a href="#cb4-3" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.scorer <span class="im">import</span> answer</span>
@@ -516,7 +516,7 @@ <h3 class="unlisted anchored" data-anchor-id="setup-1">Setup</h3>
 <section id="eval-1" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="eval-1">Eval</h3>
 <p>We’ll load the dataset from <a href="https://huggingface.co/datasets/Rowan/hellaswag">HuggingFace</a> using the <code>hf_dataset()</code> function. We’ll draw data from the validation split, and use the <code>record_to_sample()</code> function to parse the records (we’ll also pass <code>trust=True</code> to indicate that we are okay with Hugging Face executing the dataset loading code provided by hellaswag):</p>
-<div id="ff486af8" class="cell">
+<div id="110ece76" class="cell">
 <div class="sourceCode cell-code" id="cb5"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb5-1"><a href="#cb5-1" aria-hidden="true" tabindex="-1"></a><span class="at">@task</span></span>
 <span id="cb5-2"><a href="#cb5-2" aria-hidden="true" tabindex="-1"></a><span class="kw">def</span> hellaswag():</span>
 <span id="cb5-3"><a href="#cb5-3" aria-hidden="true" tabindex="-1"></a>   </span>
@@ -572,7 +572,7 @@ <h2 class="anchored" data-anchor-id="sec-theory-of-mind">Theory of Mind</h2>
 <section id="eval-2" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="eval-2">Eval</h3>
 <p>This example demonstrates adding parameters to a <code>@task</code> function to create dynamic variants of an evaluation. Here we use a <code>critique</code> parameter to determine whether a <code>self_critique()</code> solver is able to improve on the model’s baseline answer.</p>
-<div id="40588e13" class="cell">
+<div id="a3d618f6" class="cell">
 <div class="sourceCode cell-code" id="cb7"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb7-1"><a href="#cb7-1" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai <span class="im">import</span> Task, <span class="bu">eval</span>, task</span>
 <span id="cb7-2"><a href="#cb7-2" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.dataset <span class="im">import</span> example_dataset</span>
 <span id="cb7-3"><a href="#cb7-3" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.scorer <span class="im">import</span> model_graded_fact</span>
@@ -626,7 +626,7 @@ <h2 class="anchored" data-anchor-id="sec-mathematics">MATH</h2>
 <section id="setup-2" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="setup-2">Setup</h3>
 <p>We’ll start by importing the functions we need from Inspect and defining a prompt that asks the model to reason step by step and respond with its answer on a line at the end. It also nudges the model not to enclose its answer in <code>\boxed</code>, a LaTeX command for displaying equations that models often use in math output.</p>
-<div id="abfba008" class="cell">
+<div id="d891256f" class="cell">
 <div class="sourceCode cell-code" id="cb9"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb9-1"><a href="#cb9-1" aria-hidden="true" tabindex="-1"></a><span class="im">import</span> re</span>
 <span id="cb9-2"><a href="#cb9-2" aria-hidden="true" tabindex="-1"></a></span>
 <span id="cb9-3"><a href="#cb9-3" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai <span class="im">import</span> Task, task</span>
@@ -660,7 +660,7 @@ <h3 class="unlisted anchored" data-anchor-id="setup-2">Setup</h3>
 <section id="eval-3" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="eval-3">Eval</h3>
 <p>Here is the basic setup for our eval. We <code>shuffle</code> the dataset so that when we use <code>--limit</code> to develop on smaller slices we get some variety of inputs and results:</p>
-<div id="5658cdcb" class="cell">
+<div id="3ecbc34a" class="cell">
 <div class="sourceCode cell-code" id="cb10"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb10-1"><a href="#cb10-1" aria-hidden="true" tabindex="-1"></a><span class="at">@task</span></span>
 <span id="cb10-2"><a href="#cb10-2" aria-hidden="true" tabindex="-1"></a><span class="kw">def</span> math(shuffle<span class="op">=</span><span class="va">True</span>):</span>
 <span id="cb10-3"><a href="#cb10-3" aria-hidden="true" tabindex="-1"></a>    <span class="cf">return</span> Task(</span>
@@ -681,7 +681,7 @@ <h3 class="unlisted anchored" data-anchor-id="eval-3">Eval</h3>
 <span id="cb10-18"><a href="#cb10-18" aria-hidden="true" tabindex="-1"></a>    )</span></code><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></pre></div>
 </div>
 <p>The heart of this eval isn’t in the task definition though, rather it’s in how we grade the output. Math expressions can be logically equivalent but not literally the same. Consequently, we’ll use a model to assess whether the output and the target are logically equivalent. the <code>expression_equivalence()</code> custom scorer implements this:</p>
-<div id="64338a3f" class="cell">
+<div id="6dca8516" class="cell">
 <div class="sourceCode cell-code" id="cb11"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb11-1"><a href="#cb11-1" aria-hidden="true" tabindex="-1"></a><span class="at">@scorer</span>(metrics<span class="op">=</span>[accuracy(), stderr()])</span>
 <span id="cb11-2"><a href="#cb11-2" aria-hidden="true" tabindex="-1"></a><span class="kw">def</span> expression_equivalence():</span>
 <span id="cb11-3"><a href="#cb11-3" aria-hidden="true" tabindex="-1"></a>    <span class="cf">async</span> <span class="kw">def</span> score(state: TaskState, target: Target):</span>
@@ -779,7 +779,7 @@ <h2 class="anchored" data-anchor-id="sec-biology-qa">Biology QA</h2>
 <section id="eval-4" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="eval-4">Eval</h3>
 <p>Note that in the sample records above the dataset columns are not <strong>input</strong> and <strong>target</strong> so we’ll use a custom <code>FieldSpec</code> in our call to <code>example_dataset</code>. We also call the <code>use_tools()</code> function, passing <code>web_search()</code> as a tool—this gives the model access to a Google Search API that can be used to fill in background knowledge or specific facts. We use a <code>model_graded_qa()</code> scorer to more reliably score longer form model output.</p>
-<div id="d3292e28" class="cell">
+<div id="f3e3615c" class="cell">
 <div class="sourceCode cell-code" id="cb15"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb15-1"><a href="#cb15-1" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai <span class="im">import</span> Task, <span class="bu">eval</span>, task</span>
 <span id="cb15-2"><a href="#cb15-2" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.dataset <span class="im">import</span> FieldSpec, example_dataset</span>
 <span id="cb15-3"><a href="#cb15-3" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.scorer <span class="im">import</span> model_graded_qa</span>
@@ -844,7 +844,7 @@ <h2 class="anchored" data-anchor-id="sec-arc">ARC</h2>
 <section id="setup-3" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="setup-3">Setup</h3>
 <p>We’ll start by importing what we need from Inspect and writing a <code>record_to_sample()</code> function to convert raw records to samples (note that the choices and labels are encoded in JSON within the <strong>choices</strong> field so need some special pre-processing).</p>
-<div id="b6001dd3" class="cell">
+<div id="e75b93a3" class="cell">
 <div class="sourceCode cell-code" id="cb18"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb18-1"><a href="#cb18-1" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai <span class="im">import</span> Task, <span class="bu">eval</span>, task</span>
 <span id="cb18-2"><a href="#cb18-2" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.dataset <span class="im">import</span> Sample, hf_dataset</span>
 <span id="cb18-3"><a href="#cb18-3" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.scorer <span class="im">import</span> answer</span>
@@ -872,7 +872,7 @@ <h3 class="unlisted anchored" data-anchor-id="setup-3">Setup</h3>
 <section id="eval-5" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="eval-5">Eval</h3>
 <p>The ARC dataset has two subsets (ARC-Easy and ARC-Challenge). We’ll create a shared task function that can be used to run either, and then export two <code>@task</code> decorated functions so that they can be run all together or in isolation.</p>
-<div id="53b6c65d" class="cell">
+<div id="b2ee8d01" class="cell">
 <div class="sourceCode cell-code" id="cb19"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb19-1"><a href="#cb19-1" aria-hidden="true" tabindex="-1"></a><span class="kw">def</span> arc_task(dataset_name):</span>
 <span id="cb19-2"><a href="#cb19-2" aria-hidden="true" tabindex="-1"></a>   <span class="cf">return</span> Task(</span>
 <span id="cb19-3"><a href="#cb19-3" aria-hidden="true" tabindex="-1"></a>     dataset<span class="op">=</span>hf_dataset(</span>
@@ -912,7 +912,7 @@ <h2 class="anchored" data-anchor-id="sec-tool-use">Tool Use</h2>
 <section id="addition" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="addition">Addition</h3>
 <p>We’ll demonstrate with a simple tool that adds two numbers, using the <code>@tool</code> decorator to register it with the system:</p>
-<div id="121abc28" class="cell">
+<div id="44cb58e6" class="cell">
 <div class="sourceCode cell-code" id="cb21"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb21-1"><a href="#cb21-1" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai <span class="im">import</span> Task, <span class="bu">eval</span>, task</span>
 <span id="cb21-2"><a href="#cb21-2" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.dataset <span class="im">import</span> Sample</span>
 <span id="cb21-3"><a href="#cb21-3" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.scorer <span class="im">import</span> includes, match</span>
@@ -947,7 +947,7 @@ <h3 class="unlisted anchored" data-anchor-id="addition">Addition</h3>
 <span id="cb23-3"><a href="#cb23-3" aria-hidden="true" tabindex="-1"></a>    y: Second number to add.</span></code><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></pre></div>
 <p>Type annotations and descriptions are <em>required</em> for tool declarations so that the model can be informed which types to pass back to the tool function and what the purpose of each parameter is.</p>
 <p>Now that we’ve defined the tool, we can use it in an evaluation by passing it to the <code>use_tools()</code> function.</p>
-<div id="b9488377" class="cell">
+<div id="f3125461" class="cell">
 <div class="sourceCode cell-code" id="cb24"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb24-1"><a href="#cb24-1" aria-hidden="true" tabindex="-1"></a><span class="at">@task</span></span>
 <span id="cb24-2"><a href="#cb24-2" aria-hidden="true" tabindex="-1"></a><span class="kw">def</span> addition_problem():</span>
 <span id="cb24-3"><a href="#cb24-3" aria-hidden="true" tabindex="-1"></a>    <span class="cf">return</span> Task(</span>
@@ -996,7 +996,7 @@ <h3 class="unlisted anchored" data-anchor-id="setup-4">Setup</h3>
 <li><code>record_to_sample()</code> to convert raw records to samples. Note that we need a function rather than just mapping field names with a <code>FieldSpec</code> because the <strong>answer</strong> field in the dataset needs to be divided into reasoning and the actual answer (which appears at the very end after <code>####</code>).</li>
 <li><code>sample_to_fewshot()</code> to generate fewshot examples from samples.</li>
 </ol>
-<div id="ac43b81d" class="cell">
+<div id="d5f32ecf" class="cell">
 <div class="sourceCode cell-code" id="cb26"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb26-1"><a href="#cb26-1" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai <span class="im">import</span> Task, task</span>
 <span id="cb26-2"><a href="#cb26-2" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.dataset <span class="im">import</span> Sample, hf_dataset</span>
 <span id="cb26-3"><a href="#cb26-3" aria-hidden="true" tabindex="-1"></a><span class="im">from</span> inspect_ai.scorer <span class="im">import</span> match</span>
@@ -1045,7 +1045,7 @@ <h3 class="unlisted anchored" data-anchor-id="setup-4">Setup</h3>
 <section id="eval-6" class="level3 unlisted">
 <h3 class="unlisted anchored" data-anchor-id="eval-6">Eval</h3>
 <p>We’ll load the dataset from <a href="https://huggingface.co/datasets/gsm8k">HuggingFace</a> using the <code>hf_dataset()</code> function. By default we use 10 fewshot examples, but the <code>fewshot</code> task arg can be used to turn this up, down, or off. The <code>fewshot_seed</code> is provided for stability of fewshot examples across runs.</p>
-<div id="b036c923" class="cell">
+<div id="df97d2ad" class="cell">
 <div class="sourceCode cell-code" id="cb28"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb28-1"><a href="#cb28-1" aria-hidden="true" tabindex="-1"></a><span class="at">@task</span></span>
 <span id="cb28-2"><a href="#cb28-2" aria-hidden="true" tabindex="-1"></a><span class="kw">def</span> gsm8k(fewshot<span class="op">=</span><span class="dv">10</span>, fewshot_seed<span class="op">=</span><span class="dv">42</span>):</span>
 <span id="cb28-3"><a href="#cb28-3" aria-hidden="true" tabindex="-1"></a>    <span class="co"># build plan dynamically (may or may not be doing fewshot)</span></span>
diff --git a/index.html b/index.html
index 9871cff1a..c3dcc5580 100644
--- a/index.html
+++ b/index.html
@@ -1020,7 +1020,7 @@ <h2 class="anchored" data-anchor-id="learning-more">Learning More</h2>
     </div>
   </div>
 </footer>
-<script>var lightboxQuarto = GLightbox({"selector":".lightbox","descPosition":"bottom","loop":false,"closeEffect":"zoom","openEffect":"zoom"});
+<script>var lightboxQuarto = GLightbox({"descPosition":"bottom","loop":false,"selector":".lightbox","openEffect":"zoom","closeEffect":"zoom"});
 (function() {
   let previousOnload = window.onload;
   window.onload = () => {
diff --git a/log-viewer.html b/log-viewer.html
index 130e87f68..1360a4102 100644
--- a/log-viewer.html
+++ b/log-viewer.html
@@ -985,7 +985,7 @@ <h2 class="anchored" data-anchor-id="viewer-embedding">Viewer Embedding</h2>
     </div>
   </div>
 </footer>
-<script>var lightboxQuarto = GLightbox({"closeEffect":"zoom","descPosition":"bottom","loop":false,"openEffect":"zoom","selector":".lightbox"});
+<script>var lightboxQuarto = GLightbox({"closeEffect":"zoom","openEffect":"zoom","loop":false,"descPosition":"bottom","selector":".lightbox"});
 (function() {
   let previousOnload = window.onload;
   window.onload = () => {
diff --git a/search.json b/search.json
index 288cef0d8..9ac52e303 100644
--- a/search.json
+++ b/search.json
@@ -385,7 +385,7 @@
     "href": "solvers.html#built-in-solvers",
     "title": "Solvers",
     "section": "Built-In Solvers",
-    "text": "Built-In Solvers\nInspect has a number of built-in solvers, each of which can be customised in some fashion. Built in solvers can be imported from the inspect_ai.solver module. Below is a summary of these solvers. There is not (yet) reference documentation on these functions so the best way to learn about how they can be customised, etc. is to use the Go to Definition command in your source editor.\n\nsystem_message()\nPrepend role=“system” message to the list of messages (will follow any other system messages it finds in the message stream).\nprompt_template()\nModify the user prompt by substituting the current prompt into the {prompt} placeholder within the specified template. Also automatically substitutes any variables defined in sample metadata as well as any other custom named paramters passed in params.\nchain_of_thought()\nStandard chain of thought template with {prompt} substitution variable. Asks the model to provide the final answer on a line by itself at the end for easier scoring.\nuse_tools()\nDefine the set tools available for use by the model during generate().\ngenerate()\nAs illustrated above, just a simple call to generate(state). This is the default solver if no plan is specified.\nself_critique()\nPrompts the model to critique the results of a previous call to generate() (note that this need not be the same model as they one you are evaluating—use the model parameter to choose another model). Makes use of {question} and {completion} template variables. Also automatically substitutes any variables defined in sample metadata\nmultiple_choice()\nA solver which presents A,B,C,D style choices from input samples and calls generate() to yield model output. This solver should nearly always paired with the choices() scorer. Learn more about Multiple Choice in the section below.",
+    "text": "Built-In Solvers\nInspect has a number of built-in solvers, each of which can be customised in some fashion. Built in solvers can be imported from the inspect_ai.solver module. Below is a summary of these solvers. There is not (yet) reference documentation on these functions so the best way to learn about how they can be customised, etc. is to use the Go to Definition command in your source editor.\n\nsystem_message()\nPrepend role=“system” message to the list of messages (will follow any other system messages it finds in the message stream). Also automatically substitutes any variables defined in sample metadata as well as any other custom named paramters passed in params.\nprompt_template()\nModify the user prompt by substituting the current prompt into the {prompt} placeholder within the specified template. Also automatically substitutes any variables defined in sample metadata as well as any other custom named paramters passed in params.\nchain_of_thought()\nStandard chain of thought template with {prompt} substitution variable. Asks the model to provide the final answer on a line by itself at the end for easier scoring.\nuse_tools()\nDefine the set tools available for use by the model during generate().\ngenerate()\nAs illustrated above, just a simple call to generate(state). This is the default solver if no plan is specified.\nself_critique()\nPrompts the model to critique the results of a previous call to generate() (note that this need not be the same model as they one you are evaluating—use the model parameter to choose another model). Makes use of {question} and {completion} template variables. Also automatically substitutes any variables defined in sample metadata\nmultiple_choice()\nA solver which presents A,B,C,D style choices from input samples and calls generate() to yield model output. This solver should nearly always paired with the choices() scorer. Learn more about Multiple Choice in the section below.",
     "crumbs": [
       "Components",
       "<span class='chapter-number'>6</span>  <span class='chapter-title'>Solvers</span>"
diff --git a/sitemap.xml b/sitemap.xml
index 28a1ced81..2c11d8c87 100644
--- a/sitemap.xml
+++ b/sitemap.xml
@@ -18,11 +18,11 @@
   </url>
   <url>
     <loc>https://inspect.ai-safety-institute.org.uk/examples.html</loc>
-    <lastmod>2024-09-08T11:25:57.152Z</lastmod>
+    <lastmod>2024-09-08T12:15:34.237Z</lastmod>
   </url>
   <url>
     <loc>https://inspect.ai-safety-institute.org.uk/solvers.html</loc>
-    <lastmod>2024-09-07T17:45:52.516Z</lastmod>
+    <lastmod>2024-09-08T12:15:00.578Z</lastmod>
   </url>
   <url>
     <loc>https://inspect.ai-safety-institute.org.uk/tools.html</loc>
@@ -30,7 +30,7 @@
   </url>
   <url>
     <loc>https://inspect.ai-safety-institute.org.uk/agents.html</loc>
-    <lastmod>2024-09-08T11:24:31.573Z</lastmod>
+    <lastmod>2024-09-08T11:26:47.584Z</lastmod>
   </url>
   <url>
     <loc>https://inspect.ai-safety-institute.org.uk/scorers.html</loc>
diff --git a/solvers.html b/solvers.html
index 61cafda88..462fac481 100644
--- a/solvers.html
+++ b/solvers.html
@@ -419,7 +419,7 @@ <h2 class="anchored" data-anchor-id="built-in-solvers">Built-In Solvers</h2>
 <p>Inspect has a number of built-in solvers, each of which can be customised in some fashion. Built in solvers can be imported from the <code>inspect_ai.solver</code> module. Below is a summary of these solvers. There is not (yet) reference documentation on these functions so the best way to learn about how they can be customised, etc. is to use the <strong>Go to Definition</strong> command in your source editor.</p>
 <ul>
 <li><p><code>system_message()</code></p>
-<p>Prepend role=“system” <code>message</code> to the list of messages (will follow any other system messages it finds in the message stream).</p></li>
+<p>Prepend role=“system” <code>message</code> to the list of messages (will follow any other system messages it finds in the message stream). Also automatically substitutes any variables defined in sample <code>metadata</code> as well as any other custom named paramters passed in <code>params</code>.</p></li>
 <li><p><code>prompt_template()</code></p>
 <p>Modify the user prompt by substituting the current prompt into the <code>{prompt}</code> placeholder within the specified template. Also automatically substitutes any variables defined in sample <code>metadata</code> as well as any other custom named paramters passed in <code>params</code>.</p></li>
 <li><p><code>chain_of_thought()</code></p>
diff --git a/vscode.html b/vscode.html
index 89c54dc34..10f027b4e 100644
--- a/vscode.html
+++ b/vscode.html
@@ -908,7 +908,7 @@ <h2 class="anchored" data-anchor-id="troubleshooting">Troubleshooting</h2>
     </div>
   </div>
 </footer>
-<script>var lightboxQuarto = GLightbox({"loop":false,"openEffect":"zoom","selector":".lightbox","descPosition":"bottom","closeEffect":"zoom"});
+<script>var lightboxQuarto = GLightbox({"openEffect":"zoom","selector":".lightbox","closeEffect":"zoom","loop":false,"descPosition":"bottom"});
 (function() {
   let previousOnload = window.onload;
   window.onload = () => {
diff --git a/workflow.html b/workflow.html
index 37cfb457d..fa8790a65 100644
--- a/workflow.html
+++ b/workflow.html
@@ -1130,7 +1130,7 @@ <h2 class="anchored" data-anchor-id="eval-suites">Eval Suites</h2>
     </div>
   </div>
 </footer>
-<script>var lightboxQuarto = GLightbox({"closeEffect":"zoom","selector":".lightbox","descPosition":"bottom","openEffect":"zoom","loop":false});
+<script>var lightboxQuarto = GLightbox({"descPosition":"bottom","openEffect":"zoom","selector":".lightbox","closeEffect":"zoom","loop":false});
 (function() {
   let previousOnload = window.onload;
   window.onload = () => {