Top Risks

• Brand impersonation: uses 'ChatGPT' in title without verified OpenAI affiliation.
• Privacy policy is 58 chars long, fetched but not scoped to this extension — effectively a placeholder.
• Install and uninstall URL hijack both flagged; targets not confirmed benign.
• No CSP on MV3 extension; innerHTML sink in content script on chatgpt.com could be exploited.
• identity.email permission silently exposes user's Google email to developer-controlled endpoints (userapi.thinksolv.com, workers.dev).

Evidence

• brand_impersonation store brand_mention.is_impersonation=true; confirmed_owner=false; 'ChatGPT' used in title without OpenAI verification.
• privacy_policy_stub api Policy fetched but length=58 chars; scope_extension=false, data_collection=false — effectively empty/generic.
• install_url_hijack crx install_url_hijack=true, uninstall_url_hijack=true; targets null (not verified safe).
• dom_xss_sink_no_csp crx innerHTML from variable in content.bundle.js; csp_present=false increases exploitability.
• identity_email_with_external_endpoints crx identity.email + calls to userapi.thinksolv.com and aruchamyvikram.workers.dev — email tied to external infra.
• no_csp_mv3 manifest content_security_policy=null; MV3 default applies but no explicit CSP restricts external script loading.
• external_host_diversity crx 12 distinct external JS hosts across CA/TW/US; includes Cloudflare Workers under dev's personal subdomain.
• no_developer_name store developer_name is empty string; no 'Offered by' display name; not a verified publisher.

Permissions Breakdown

• downloads medium Allows saving files to disk; expected for Word/PDF export function.
• identity medium OAuth token access; can authenticate user silently to Google services.
• identity.email medium Exposes user's Google account email to the extension.
• storage low Local/sync storage for settings; low direct harm.
• https://chatgpt.com/* medium Content script on chatgpt.com only; narrowly scoped but reads all page content.

Pillar Scores

Bookkeeping