Fix page excerpt handling of script and style closing tags

CHANGELOG.md

@@ -5,6 +5,13 @@ the section in the pull request that bumps the version: a one-line summary
 paragraph, then one bullet per change a user or operator would notice. Keep
 each paragraph and bullet on one line; release pages render line breaks.
 
+## 0.19.1 - 2026-10-01
+
+Page claim checks exclude script and stylesheet bodies with HTML closing-tag whitespace or trailing attributes.
+
+- Keep hidden script and style text out of page evidence when closing tags contain HTML whitespace or trailing attributes, or the fetched page ends inside a raw-text body.
+- Preserve visible text in similarly named custom elements and retain existing fetch and evidence limits.
+
 ## 0.19.0 - 2026-09-30
 
 Sys1 can update supported global CLI installations before work starts.

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hraness/sys1",
-  "version": "0.19.0",
+  "version": "0.19.1",
   "description": "Sys1 gives coding agents tools to review code, check completion claims, and get structured answers from Jev or a local model.",
   "type": "module",
   "license": "MIT",

src/verify/evidence.ts

@@ -180,7 +180,10 @@ export async function fetchPage(
       }
       const text = new TextDecoder().decode(whole);
       evidence.ok = true;
-      evidence.excerpt = text.replace(/<script[\s\S]*?<\/script>/gi, " ").replace(/<style[\s\S]*?<\/style>/gi, " ")
+      // Raw-text bodies are not visible evidence. Accept HTML close-tag whitespace
+      // and discard an unfinished body when the bounded page ends inside it.
+      evidence.excerpt = text.replace(/<script(?=[\t\n\f\r />])[^>]*>[\s\S]*?(?:<\/script(?=[\t\n\f\r />])[^>]*>|$)/gi, " ")
+        .replace(/<style(?=[\t\n\f\r />])[^>]*>[\s\S]*?(?:<\/style(?=[\t\n\f\r />])[^>]*>|$)/gi, " ")
         .replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim().slice(0, EVIDENCE_LIMITS.maxPageBytes);
       return evidence;
     } catch {

src/version.ts

@@ -1 +1 @@
-export const SYS1_VERSION = "0.19.0";
+export const SYS1_VERSION = "0.19.1";

test/verify-page-excerpt.test.ts

@@ -0,0 +1,35 @@
+import { expect, test } from "bun:test";
+import { EVIDENCE_LIMITS, fetchPage } from "../src/verify/evidence.ts";
+
+for (const tag of ["script", "style"]) {
+  for (const whitespace of [" ", "\t", "\n", "\r", "\f", " \n\t"]) {
+    test(`page excerpt excludes ${tag} with close-tag whitespace ${JSON.stringify(whitespace)}`, async () => {
+      const page = await fetchPage("https://example.com", async () => new Response(
+        `<p>Visible before.</p><${tag.toUpperCase()} type="text/plain">hidden-evidence-canary</${tag}${whitespace}><p>Visible after.</p>`,
+      ));
+      expect(page.ok).toBe(true);
+      expect(page.excerpt).toBe("Visible before. Visible after.");
+    });
+  }
+  for (const tail of ["/", " ignored", " ignored=\"value\""]) {
+    test(`page excerpt retains following text after ${tag} end-tag tail ${tail}`, async () => {
+      const page = await fetchPage("https://example.com", async () => new Response(
+        `<p>Before.</p><${tag}>hidden-evidence-canary</${tag}${tail}><p>After.</p>`,
+      ));
+      expect(page.excerpt).toBe("Before. After.");
+    });
+  }
+  test(`page excerpt excludes ${tag} body cut off by the byte cap`, async () => {
+    const page = await fetchPage("https://example.com", async () => new Response(
+      `<p>Visible.</p><${tag}>` + "hidden-evidence-canary ".repeat(EVIDENCE_LIMITS.maxPageBytes) + `</${tag}>`,
+    ));
+    expect(page.excerpt).toBe("Visible.");
+  });
+}
+
+test("similarly named custom elements retain their visible text", async () => {
+  const page = await fetchPage("https://example.com", async () => new Response(
+    "<script-example>Visible example.</script-example><style-guide>Visible guide.</style-guide>",
+  ));
+  expect(page.excerpt).toBe("Visible example. Visible guide.");
+});