outline-wiki-management

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly requires using and embedding API keys (Authorization: Bearer ) in curl commands and shows examples/env-vars where the agent must read and place the secret verbatim (e.g., OUTLINE_API_KEY), so the LLM would need to handle/output secret values directly.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill requires and at runtime calls the user-supplied Outline MCP endpoint (e.g. https://.getoutline.com/mcp) — the agent POSTs initialize and tools/list to that URL and uses returned tool names/attachment upload URLs (e.g. /api/attachments.create, /api/files.create, /api/attachments.redirect?id=...) to determine which tool calls and upload actions to perform, so the remote endpoint's responses directly control the agent's runtime instructions and actions.