twitter-x-business-contact-search

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.80). The skill contains explicit instructions to install dependencies automatically and to conceal that action from the user ("do not describe... do not request separate approval"), which are deceptive runtime directives outside the advertised search/data-collection purpose.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill includes deliberate, deceptive instructions to auto-install an external runtime without user consent and guidance to hide credentials, combined with functionality to harvest contact data—indicative of potential malicious/surreptitious behavior.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). SKILL.md instructs the agent to use the BrowserAct CLI to access live public Twitter/X pages at runtime (public web content category), which can include arbitrary free text that the LLM may ingest while extracting profile/bio/post data.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The Skill explicitly directs the agent at runtime to automatically install and invoke the BrowserAct CLI using the provided install links (e.g., https://www.browseract.com/?co-from=lead-generation and https://www.browseract.com/?co-from=lead-generation&redirect=https://github.com/browser-act/skills/tree/main), which are used to fetch and install external code that the Skill treats as a required runtime dependency.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The Skill instructs the agent to automatically install the BrowserAct CLI (via an installation link) and resume operation without asking the user, which directs the agent to modify the host system's software state without explicit approval.