Top Risks

• Privacy policy URL returns error (fetched==false); effective policy unknown — scored +10.0.
• Free-webmail dev (gmail) with no verified business; brand_mention flags GPT-4/Claude impersonation.
• <all_urls> host permission with content_scripts on every page enables broad data access.
• AI extension contacts api.anthropic.com and api.openai.com; page content may be transmitted to external AI APIs.
• Only 11 installs with HIGH-tier permission (small_install_high_perm) raises tail-attack-surface concern.

Evidence

• privacy_policy_fetch_failed api Privacy policy URL returns HTTPError; policy cannot be evaluated — worst-case score applied.
• free_webmail_dev store Developer email deepak.chowdary147@gmail.com; no verified business domain; not verified publisher.
• brand_impersonation store brand_mention.is_impersonation=true; brands gpt-4 and claude mentioned; confirmed_owner=false.
• host_permission_all_urls manifest <all_urls> host permission + content_scripts on *://*/* grants read/write on every page.
• ai_page_content_exfil manifest AI translation extension contacts api.anthropic.com and api.openai.com; page image/text sent externally.
• small_install_high_perm store Only 11 installs with HIGH-tier permission; anomaly flag raised by install_perm_anomaly.
• no_csp manifest content_security_policy is null; MV3 default CSP applies but no explicit hardening declared.
• cve_none crx cve_findings_raw is empty; no bundled vulnerable libraries detected.

Permissions Breakdown

• activeTab low Access only to the currently active tab on user action; scoped and low risk.
• storage low Local storage for settings/state; no exfil risk alone.
• <all_urls> (host_permission) high Grants content script injection on every site; broad capability paired with AI processing.

Pillar Scores

Bookkeeping