build(deps): bump actions/checkout from 4 to 7

Bumps [actions/checkout](https://github.com/actions/checkout) from 4 to 7.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/checkout/compare/v4...v7)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-version: '7'
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

.github/dependabot.yml

@@ -0,0 +1,10 @@
+version: 2
+updates:
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "weekly"
+    open-pull-requests-limit: 5
+    groups:
+      actions-minor:
+        update-types: ["minor", "patch"]

.github/workflows/ci.yml

@@ -9,7 +9,7 @@ jobs:
   validate:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v7
 
       - name: Validate manifest JSON
         run: |

.github/workflows/release.yml

@@ -18,7 +18,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
         with:
           ref: ${{ github.event.inputs.tag || github.ref }}
 

CHANGELOG.md

@@ -0,0 +1,58 @@
+# Changelog
+
+All notable changes to KeyFinder are documented here. Format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/). Versioning follows [SemVer](https://semver.org/spec/v2.0.0.html).
+
+## [2.1.1] - 2026-05-14
+
+### Added
+- `SECURITY.md` with threat model, disclosure policy, and known limitations of the MAIN <-> ISOLATED nonce bridge
+- `.github/dependabot.yml` for weekly GitHub Actions version bumps
+- `CHANGELOG.md`
+
+### Changed
+- CSV export sanitiser now also prefixes cells starting with LF (`\n`), not just `=`, `+`, `-`, `@`, tab, CR
+- Popup and results page version label is now read from the manifest at runtime instead of being hardcoded
+
+### Fixed
+- Window-global scan in `js/interceptor.js` now runs at `document_start`, `DOMContentLoaded`, and `load`, with per-name dedupe. The previous implementation only scanned at `document_start` when page globals had not yet been assigned, making the entire pass dead code on most real pages
+
+## [2.1.0] - 2026-04-14
+
+### Added
+- Per-session nonce validation between MAIN-world interceptor and ISOLATED content script to prevent forged finding injection
+- CSV formula-injection sanitiser on findings export
+- Serialised storage writes to eliminate cross-tab race conditions
+- 5000-finding cap with FIFO eviction
+- Per-tab alert badge with red-dot icon overlay when secrets are detected
+- MutationObserver scans dynamically-injected DOM nodes for SPA coverage
+- Explicit Content Security Policy in Chrome and Firefox manifests
+- `js/interceptor-loader.js` for both browsers, replacing direct MAIN-world content script on Firefox so the nonce handoff actually works
+- GitHub Actions release pipeline (`.github/workflows/release.yml`): on `v*` tag, build Chrome + Firefox zips, compute SHA256, attach to GitHub Release
+- GitHub Actions CI pipeline (`.github/workflows/ci.yml`): manifest JSON validation, Chrome <-> Firefox version parity check, build verification, `web-ext lint` on the Firefox bundle
+
+### Changed
+- Keyword input validation: 50 character maximum, 50 keyword maximum
+- Findings are now deleted by unique ID instead of URL substring match
+- URL parameter scanner uses exact match instead of substring (was matching `author` as `auth`)
+- Keyword scanner enforces word boundaries (was matching `key` inside `hotkey`, `monkey`)
+- camelCase JS identifiers are now skipped in keyword value matches
+- Sentry DSN downgraded from `high` to `low` severity (public by design)
+
+### Fixed
+- Stored finding race conditions across concurrent tabs
+- False positives from GitHub localStorage caches (`ref-selector:*`, `jump_to:*`, `soft-nav:*`, `COPILOT_*`)
+- False positives from common CSRF tokens (`authenticity_token`, `csrf_token`, `__RequestVerificationToken`)
+- False positives from keyboard shortcut data attributes (`data-hotkey`, `data-hotkey-scope`)
+
+## [2.0.0] - 2026-04-07
+
+### Added
+- Complete rewrite to Manifest V3
+- Enterprise-grade secret detection with 80+ regex patterns covering AWS, GCP, Azure, GitHub, GitLab, Stripe, PayPal, Square, Slack, Discord, and more
+- Firefox support (MV3, Firefox 128+)
+- Privacy policy
+- Replaced demo gifs with professional logo
+
+### Removed
+- Manifest V2 background page
+- Legacy jQuery dependency

README.md

@@ -60,10 +60,13 @@ Additionally, **Shannon entropy analysis** is applied to detect random high-entr
 - **Zero dependencies** - Pure vanilla JavaScript, no jQuery, no external libraries
 - **Manifest V3** - Built for modern Chrome and Firefox with service worker architecture
 - **Passive scanning** - Runs automatically on every page load
-- **Custom keywords** - Add your own search terms to scan for
+- **SPA-aware** - MutationObserver re-scans dynamically injected DOM
+- **Per-tab alert badge** - Red-dot icon overlay when a tab has findings
+- **Custom keywords** - Add your own search terms to scan for (validated, 50 max)
 - **Dashboard** - Professional results page with filtering, sorting, and search
-- **Export** - Download findings as JSON or CSV
-- **Badge counter** - Shows finding count on the extension icon
+- **Export** - Download findings as JSON or CSV (with formula-injection sanitiser)
+- **Hardened bridge** - Per-page nonce on MAIN <-> ISOLATED CustomEvent channel
+- **Bounded storage** - 5000-finding cap with FIFO eviction; serialised writes across tabs
 - **Low footprint** - Minimal CPU and memory usage
 - **All frames** - Scans iframes and embedded content
 
@@ -117,8 +120,9 @@ keyFinder/
   js/
     background.js          # Service worker - storage and message handling
     patterns.js            # 80+ secret detection regex patterns
-    content.js             # Page scanner - DOM, scripts, network interception
-    interceptor.js         # XHR/Fetch hooking and window global scanning
+    content.js             # ISOLATED-world page scanner - DOM, scripts, network
+    interceptor-loader.js  # ISOLATED loader - sets nonce, injects MAIN-world interceptor
+    interceptor.js         # MAIN-world XHR/Fetch hooks + window global scanning
     popup.js               # Popup logic
     results.js             # Dashboard logic with filtering and export
   css/
@@ -132,10 +136,22 @@ keyFinder/
     build.sh               # Build Chrome and Firefox zip packages
 ```
 
+## Security
+
+See [SECURITY.md](SECURITY.md) for the threat model, disclosure policy, and known limitations. Release notes are in [CHANGELOG.md](CHANGELOG.md).
+
 ## Disclaimer
 
 This tool is intended for **security research and authorized testing only**. Use it to identify leaked secrets on your own applications or during authorized penetration tests. You are responsible for your own actions.
 
+
+## Professional services
+
+keyFinder is built and maintained by [GreyCore Labs](https://greycorelabs.com), a US-incorporated offensive security firm. Want the same eye on your own product?
+
+- [Penetration testing](https://greycorelabs.com/#plans) - web, API, mobile, cloud. Fixed quote within 24 hours, redacted sample report on request.
+- [Free external attack-surface scan](https://greycorelabs.com/#free-scan) - one-page report in 48 hours, no strings attached.
+
 ## License
 
 MIT

SECURITY.md

@@ -0,0 +1,57 @@
+# Security Policy
+
+## Reporting a vulnerability
+
+Email **security reports** privately to the address on the maintainer's GitHub profile.
+Do **not** open a public issue for unpatched vulnerabilities.
+
+When reporting, include:
+- Affected version (`manifest.json` `version` field)
+- Browser and version
+- Steps to reproduce
+- Impact assessment (what an attacker gains)
+
+A response is targeted within 7 days.
+
+## Threat model
+
+KeyFinder runs as a content script in every page the user visits and reports findings to a service worker. It is **client-side, passive, and read-only** with respect to the page.
+
+### In scope
+- Privilege escalation from a malicious page into the extension's service worker
+- Persistent storage poisoning via crafted findings
+- Cross-tab data leakage through `chrome.storage`
+- CSV / JSON export injection (formulas, embedded HTML, JS)
+- Manifest / CSP weaknesses enabling code injection into the extension's own pages
+- Pattern-rule false positives that consistently leak benign data into findings
+
+### Out of scope
+- A malicious page generating **fake findings** in the user's results view. The MAIN-world interceptor and the ISOLATED content script communicate over `CustomEvent` with a per-page nonce stored as a `data-kf-verify` attribute on `documentElement`. The nonce is removed once the content script consumes it, but a page script that runs between `document_start` and `document_idle` can read the attribute and forge events. Mitigation cost is high (Symbols don't cross realms; postMessage is also page-visible). The impact is limited to **showing the user a finding that isn't real** - no data is exfiltrated, no privileged API is reached. Treat findings on a hostile page as advisory.
+- Detection accuracy of individual regex rules. False positives and false negatives are expected; report a tuning issue rather than a CVE.
+- Extension being uninstalled or disabled by the user.
+
+## What the extension can see
+
+| Surface | Scope |
+|---|---|
+| Page DOM | All pages (`<all_urls>`) at `document_idle` (ISOLATED world) and `document_start` (MAIN world interceptor) |
+| Network | `fetch` and `XMLHttpRequest` responses initiated by the page (response bodies up to 500 KB are scanned) |
+| Web storage | `localStorage`, `sessionStorage`, `document.cookie` on the page |
+| Inter-extension | None. No host permissions beyond `activeTab` and `storage` |
+
+The extension makes **no outbound network requests** other than fetching same-origin scripts already referenced by the page.
+
+## Known limitations
+
+- **Per-tab badge dot** is set on the first finding only; subsequent findings update the count but not the icon
+- **5000 finding cap** with FIFO eviction. High-volume scans (heavy SPAs over a long session) will drop oldest findings
+- **CSV export** prefixes a single-quote on cells starting with `=`, `+`, `-`, `@`, tab, or carriage return to neutralise Excel / Sheets formula injection. Line-feed prefix is not currently neutralised
+- **Service worker restarts** drop the in-flight `storageQueue` Promise chain. Subsequent storage writes are still serialised; only pending writes from the killed worker are lost
+
+## Supported versions
+
+| Version | Supported |
+|---|---|
+| 2.1.x | yes |
+| 2.0.x | no |
+| < 2.0 | no |

js/interceptor.js

@@ -39,23 +39,39 @@
     "__ENV__", "__CONFIG__", "ENV", "CONFIG",
   ];
 
-  for (const name of globalNames) {
-    try {
-      const val = window[name];
-      if (val === undefined || val === null) continue;
-      const str = typeof val === "object" ? JSON.stringify(val) : String(val);
-      if (str.length < 8 || str === "[object Object]") continue;
-      emit({
-        match: `window.${name}=${str.substring(0, 200)}`,
-        type: "window-global",
-        patternName: "Exposed Global Variable",
-        severity: "high",
-        confidence: typeof val !== "object" && isHighEntropy(str.substring(0, 60)) ? "high" : "medium",
-        provider: "JS Global Scan",
-        isObject: typeof val === "object",
-        rawText: typeof val === "object" ? str.substring(0, 5000) : null,
-      });
-    } catch {}
+  const scannedGlobals = new Set();
+  function scanGlobals() {
+    for (const name of globalNames) {
+      if (scannedGlobals.has(name)) continue;
+      try {
+        const val = window[name];
+        if (val === undefined || val === null) continue;
+        const str = typeof val === "object" ? JSON.stringify(val) : String(val);
+        if (str.length < 8 || str === "[object Object]") continue;
+        scannedGlobals.add(name);
+        emit({
+          match: `window.${name}=${str.substring(0, 200)}`,
+          type: "window-global",
+          patternName: "Exposed Global Variable",
+          severity: "high",
+          confidence: typeof val !== "object" && isHighEntropy(str.substring(0, 60)) ? "high" : "medium",
+          provider: "JS Global Scan",
+          isObject: typeof val === "object",
+          rawText: typeof val === "object" ? str.substring(0, 5000) : null,
+        });
+      } catch {}
+    }
+  }
+  // Run at document_start (likely empty), DOMContentLoaded, and load to catch
+  // globals set at any phase. Each name reports at most once via scannedGlobals.
+  scanGlobals();
+  if (document.readyState === "loading") {
+    document.addEventListener("DOMContentLoaded", scanGlobals, { once: true });
+  }
+  if (document.readyState !== "complete") {
+    window.addEventListener("load", scanGlobals, { once: true });
+  } else {
+    scanGlobals();
   }
 
   const origXhrOpen = XMLHttpRequest.prototype.open;

js/popup.js

@@ -1,6 +1,8 @@
 document.addEventListener("DOMContentLoaded", init);
 
 async function init() {
+  const versionLabel = document.getElementById("versionLabel");
+  if (versionLabel) versionLabel.textContent = "v" + chrome.runtime.getManifest().version;
   await renderKeywords();
   await renderStats();
   document.getElementById("keywordForm").addEventListener("submit", handleAddKeyword);

js/results.js

@@ -3,6 +3,8 @@ let allFindings = [];
 document.addEventListener("DOMContentLoaded", init);
 
 async function init() {
+  const versionLabel = document.getElementById("versionLabel");
+  if (versionLabel) versionLabel.textContent = "v" + chrome.runtime.getManifest().version;
   const response = await chrome.runtime.sendMessage({ type: "getFindings" });
   allFindings = response.findings || [];
 
@@ -214,7 +216,7 @@ function exportJson() {
 
 function csvSafe(value) {
   let str = String(value || "");
-  if (/^[=+\-@\t\r]/.test(str)) str = "'" + str;
+  if (/^[=+\-@\t\r\n]/.test(str)) str = "'" + str;
   str = str.replace(/"/g, '""');
   return `"${str}"`;
 }

manifest.firefox.json

@@ -1,7 +1,7 @@
 {
   "name": "KeyFinder",
   "description": "Passively discovers API keys, tokens, and secrets leaked in page scripts, DOM, network responses, and browser storage. Available for Chrome and Firefox.",
-  "version": "2.1.0",
+  "version": "2.1.1",
   "manifest_version": 3,
   "browser_specific_settings": {
     "gecko": {

manifest.json

@@ -1,7 +1,7 @@
 {
   "name": "KeyFinder",
   "description": "Passively discovers API keys, tokens, and secrets leaked in page scripts, DOM, network responses, and browser storage. Available for Chrome and Firefox.",
-  "version": "2.1.0",
+  "version": "2.1.1",
   "manifest_version": 3,
   "action": {
     "default_icon": {

popup.html

@@ -11,7 +11,7 @@
     <div class="header-brand">
       <img src="icons/icon48.png" alt="KeyFinder" class="header-icon">
       <h1>KeyFinder</h1>
-      <span class="version">v2.0</span>
+      <span class="version" id="versionLabel"></span>
     </div>
     <p class="header-tagline">Passive API key & secret discovery</p>
   </header>

results.html

@@ -10,7 +10,7 @@
   <header class="header">
     <div class="header-left">
       <img src="icons/icon48.png" alt="KeyFinder" class="header-icon">
-      <h1>KeyFinder <span class="version">v2.0</span></h1>
+      <h1>KeyFinder <span class="version" id="versionLabel"></span></h1>
     </div>
     <div class="header-actions">
       <div class="filter-group">