Fixed the Confluence search missing relevant pages (#1021)

app/MindWork AI Studio/Tools/ToolCallingSystem/ToolCallingImplementations/ConfluenceSearchTool.cs

@@ -261,7 +261,14 @@ public sealed class ConfluenceSearchTool(WebPageRetrievalService webPageRetrieva
 
     internal static Uri BuildSearchUrl(Uri baseUrl, string query, string? spaceKey)
     {
-        var cql = $"text ~ \"{EscapeCqlValue(query)}\"";
+        //
+        // siteSearch is the field Confluence's own search box sends to this page: it finds the
+        // pages holding any of the words and ranks them by relevance. The text field, the one
+        // Atlassian documents, requires every word on the page and so missed pages the search
+        // box finds. Atlassian's CQL field reference does not list siteSearch, but the search
+        // page itself depends on it, and the REST API documentation uses it in its examples.
+        //
+        var cql = $"siteSearch ~ \"{EscapeCqlValue(query)}\"";
         if (!string.IsNullOrWhiteSpace(spaceKey))
             cql += $" and space=\"{EscapeCqlValue(spaceKey)}\"";
 

app/MindWork AI Studio/wwwroot/changelog/v26.10.1.md

@@ -9,3 +9,4 @@
 - Fixed a warning about the confidence level that stayed in the dialog for adding a local data source, even after you chose an embedding provider that meets it. The dialog now checks again whenever you change either of them, so a warning also appears when you switch to a provider that falls short.
 - Fixed two confidence levels sharing the same name in the German user interface. Moderate is called "Mäßig" again, and Medium stays "Mittel".
 - Fixed renaming an external data source (ERI server) leaving its secret behind under the old name in the keychain of your operating system.
+- Fixed the Search Confluence tool missing pages that the search of your Confluence wiki finds. The tool now searches the same way as the search box of your wiki.

app/Tests/Tools/ToolCalling/ConfluenceSearchToolTests.cs

@@ -51,14 +51,14 @@ public sealed class ConfluenceSearchToolTests
         var parameters = HttpUtility.ParseQueryString(searchUrl.Query);
 
         Assert.That(searchUrl.GetLeftPart(UriPartial.Path), Is.EqualTo("https://wiki.example.org/confluence/dosearchsite.action"), "The search page lies below the context path of the wiki, even when the configured address lacks the final slash.");
-        Assert.That(parameters["cql"], Is.EqualTo("text ~ \"release plan\""));
+        Assert.That(parameters["cql"], Is.EqualTo("siteSearch ~ \"release plan\""), "The search box of the wiki sends the same field, so the tool finds the pages the user would find there.");
         Assert.That(parameters["queryString"], Is.EqualTo("release plan"), "Confluence shows these words in its search field, so the page reads like a search the user made.");
         Assert.That(ConfluenceSearchTool.IsWithinWiki(WIKI, searchUrl), Is.True);
     }
 
-    [TestCase(@"plan"" or space = ""HR", @"text ~ ""plan\"" or space = \""HR""")]
-    [TestCase(@"C:\temp\", @"text ~ ""C:\\temp\\""")]
-    [TestCase(@"plan\"" or space = \""HR", @"text ~ ""plan\\\"" or space = \\\""HR""")]
+    [TestCase(@"plan"" or space = ""HR", @"siteSearch ~ ""plan\"" or space = \""HR""")]
+    [TestCase(@"C:\temp\", @"siteSearch ~ ""C:\\temp\\""")]
+    [TestCase(@"plan\"" or space = \""HR", @"siteSearch ~ ""plan\\\"" or space = \\\""HR""")]
     public void TheQueryCannotLeaveItsCqlString(string query, string expectedCql)
     {
         //
@@ -72,13 +72,13 @@ public sealed class ConfluenceSearchToolTests
     [Test]
     public void ASpaceKeyRestrictsTheSearchToThatSpace()
     {
-        Assert.That(Cql(ConfluenceSearchTool.BuildSearchUrl(WIKI, "release plan", "DEV")), Is.EqualTo("text ~ \"release plan\" and space=\"DEV\""));
+        Assert.That(Cql(ConfluenceSearchTool.BuildSearchUrl(WIKI, "release plan", "DEV")), Is.EqualTo("siteSearch ~ \"release plan\" and space=\"DEV\""));
     }
 
     [Test]
     public void ASpaceKeyCannotLeaveItsCqlStringEither()
     {
-        Assert.That(Cql(ConfluenceSearchTool.BuildSearchUrl(WIKI, "release plan", @"DEV"" or space = ""HR")), Is.EqualTo(@"text ~ ""release plan"" and space=""DEV\"" or space = \""HR"""), "The space key comes from the model as well, so it gets the same escaping as the words.");
+        Assert.That(Cql(ConfluenceSearchTool.BuildSearchUrl(WIKI, "release plan", @"DEV"" or space = ""HR")), Is.EqualTo(@"siteSearch ~ ""release plan"" and space=""DEV\"" or space = \""HR"""), "The space key comes from the model as well, so it gets the same escaping as the words.");
     }
 
     [TestCase(null)]
@@ -86,7 +86,7 @@ public sealed class ConfluenceSearchToolTests
     [TestCase("   ")]
     public void WithoutASpaceKeyTheWholeWikiIsSearched(string? spaceKey)
     {
-        Assert.That(Cql(ConfluenceSearchTool.BuildSearchUrl(WIKI, "release plan", spaceKey)), Is.EqualTo("text ~ \"release plan\""), "An empty space key would otherwise ask for a space which does not exist and find nothing.");
+        Assert.That(Cql(ConfluenceSearchTool.BuildSearchUrl(WIKI, "release plan", spaceKey)), Is.EqualTo("siteSearch ~ \"release plan\""), "An empty space key would otherwise ask for a space which does not exist and find nothing.");
     }
 
     [Test]

documentation/Tools.md

@@ -155,7 +155,7 @@ What differs between callers is which targets are acceptable, and that follows f
 
 `read_web_page` remains the independent single-URL tool and may use its configured private-host allowlist and operating-system sign-in behavior for allowed HTTPS targets. An allowed private host can only be read by a High-confidence provider.
 
-`search_confluence` builds a CQL query for the configured HTTPS Confluence Data Center site's `dosearchsite.action` page and loads it through `WebPageRetrievalService`, the same reader used by `read_web_page`. The model supplies a search phrase and optionally a space key, never a URL or CQL expression. The tool returns the extracted search page as Markdown with links, after truncation and prompt-injection filtering, and lists the search page as its source. Every request, redirects included, must stay within the configured base URL; `WebPageRetrievalOptions.IsTargetAllowed` refuses a redirect before it is followed, so the query never reaches another host. The operating-system sign-in goes to the configured host only when all its addresses are private, the same rule `read_web_page` follows, and a redirect to Confluence's login page is reported as a missing sign-in instead of an empty search. The tool is offered to High-confidence providers only and checks that again before each search, because a lowered tool setting must not let internal wiki content reach a less trusted provider; the result raises the chat's continuing confidence requirement to High. Selecting `search_confluence` also selects `read_web_page` so the model can load a result's full content; the latter tool's private-host allowlist and other availability rules still apply, so a wiki with a private address has to be in that allowlist before any result opens.
+`search_confluence` builds a CQL query for the configured HTTPS Confluence Data Center site's `dosearchsite.action` page and loads it through `WebPageRetrievalService`, the same reader used by `read_web_page`. The model supplies a search phrase and optionally a space key, never a URL or CQL expression. The query uses the `siteSearch` field, which Confluence's own search box sends: it finds pages holding any of the words and ranks them by relevance, while the documented `text` field requires all of them and misses pages the search box finds. The tool returns the extracted search page as Markdown with links, after truncation and prompt-injection filtering, and lists the search page as its source. Every request, redirects included, must stay within the configured base URL; `WebPageRetrievalOptions.IsTargetAllowed` refuses a redirect before it is followed, so the query never reaches another host. The operating-system sign-in goes to the configured host only when all its addresses are private, the same rule `read_web_page` follows, and a redirect to Confluence's login page is reported as a missing sign-in instead of an empty search. The tool is offered to High-confidence providers only and checks that again before each search, because a lowered tool setting must not let internal wiki content reach a less trusted provider; the result raises the chat's continuing confidence requirement to High. Selecting `search_confluence` also selects `read_web_page` so the model can load a result's full content; the latter tool's private-host allowlist and other availability rules still apply, so a wiki with a private address has to be in that allowlist before any result opens.
 
 Confluence Cloud is not supported yet. It offers neither `dosearchsite.action` as a server-rendered page nor the operating-system sign-in; its search needs Confluence's REST API with an API token instead, which is also the way to stop depending on the HTML of the Data Center search page.