extension-posting-to-x

Fail

Audited by Gen Agent Trust Hub on Oct 30, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONSAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references and fetches the x-client package from the mops.one registry. Automated security scans have flagged the URL https://mops.one/x-client as potentially malicious (Phishing). While this domain is the standard registry for the Motoko ecosystem, users should exercise caution.
  • [REMOTE_CODE_EXECUTION]: The skill instructs the user to download and install external code using the command mops add x-client@0.5.1.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text for tweets which is then sent to an external API sink.
  • Ingestion points: The body parameter in the tweet function located in src/backend/mixins/x-posting.mo.
  • Boundary markers: Absent. The raw text is used directly in the API request.
  • Capability inventory: The skill performs network outcalls to the X API via TweetsApi.createPosts in src/backend/lib/x.mo.
  • Sanitization: Absent. The input text is not sanitized before being sent to the X API.
  • [SAFE]: The skill demonstrates strong security practices for handling sensitive credentials. It utilizes per-user OAuth tokens stored in private canister memory, strictly forbids the use of getters or logs for tokens, and mandates non-replicated HTTP outcalls (is_replicated = ?false) to prevent token exposure and ensure execution determinism.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Oct 30, 2026, 08:58 PM