overleaf

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs users to copy their Overleaf session cookie and pass it verbatim on the command line (e.g., olcli auth --cookie "YOUR_SESSION_COOKIE"), which requires handling and outputting a secret value directly and is high-risk for exfiltration.

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

• Third-party content exposure detected (high risk: 0.70). The required workflow involves the MCP server returning OVERLEAF-sourced comment/project data as JSON text over stdio to the agent (i.e., outsider-authored Overleaf project comments/messages are included in the MCP tool outputs that an LLM consumes as context), which creates indirect prompt-injection risk.

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

• Potentially malicious external URL detected (high risk: 0.90). The MCP docs instruct AI agents to run npx to fetch/execute the npm package @aloth/olcli-mcp at runtime (e.g., "npx -y @aloth/olcli-mcp"), which causes remote code to be fetched and executed to start an MCP server that exposes tools to the agent.