Chat Memo | Auto-save AI Chats from ChatGPT, Gemini, Claude +
memnnheiikbfdcobfkghhfihnegkfici
Risk Score
5.34
Risk Level:
Medium
Recommendation:
🟡 MEDIUM RISK — review
Top Risks
- Privacy policy is Google's own policy — not scoped to this extension; collects and shares data with third parties.
- Brand impersonation: names ChatGPT, Gemini, Claude in title without confirmed ownership, not verified publisher for those brands.
- Install-URL hijack: onInstalled opens chatmemo.ai/welcome, redirecting users to developer's site.
- No CSP declared (MV3 strict-default mitigates, but innerHTML sinks and new Function() in jszip raise XSS risk from AI page content.
- <all_urls> content scripts extend beyond the stated AI-chat platforms, creating broad page-access surface.
Evidence
- privacy_policy_generic store Privacy URL points to Google's own policy; scope_extension=false, data_collection=true, third_party_sharing=true → worst-case privacy score.
- brand_impersonation store brand_mention.is_impersonation=true; title references ChatGPT/Gemini/Claude; confirmed_owner=false; verified_publisher=true but not for those brands.
- install_url_hijack crx install_url_hijack=true; target=https://chatmemo.ai/welcome — opens dev site on every install.
- dom_xss_sinks crx Two innerHTML-from-variable sinks in popup.js and content_common.js; no CSP; could render attacker-controlled AI chat content.
- function_constructor_in_jszip crx new Function() in lib/jszip.min.js; in bundled library context, elevated risk with no CSP.
- analytics_endpoint crx chatmemo-analytics.kylis.workers.dev in js_external_hosts; Cloudflare Worker analytics endpoint not documented in privacy policy.
- host_permissions_all_urls manifest <all_urls> host permission broader than the ~10 AI-chat domains listed in content_scripts_matches.
- developer_name_empty store developer_name is empty string; no 'Offered by' identity despite verified_publisher=true.
Permissions Breakdown
- storage low Needed to persist saved chat data locally.
- tabs medium Can read tab URLs and titles across all open tabs.
- downloads medium Can trigger file downloads; export function but also potential for abuse.
- alarms low Background scheduling; low standalone risk.
- <all_urls> (host_permissions) high Content scripts on <all_urls> gives read/write access to all pages, broader than AI-chat sites.
Pillar Scores
Permissions6.50
Reputation5.00
Network4.00
Webstore6.50
Maintenance0.00
Privacy10.00
Code Quality5.00
CVE Exposure0.00
Bookkeeping
Rubric v3.6
Scored at 2026-06-16 07:54
Listing SHA
906942ff7b12…
Force block
— not fired
Score recovered
no
Elapsed
27.8s