picoclaw

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly supports installing skills from arbitrary GitHub repo slugs and registry endpoints (references/skills.md) and configuring web search/fetch tools (references/tools.md), which causes the agent to fetch and load untrusted, user-generated third‑party content that it will read and act on as part of its workflow.

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

• Potentially malicious external URL detected (high risk: 0.80). The docs show that skills can be installed at runtime from GitHub (e.g., the command "picoclaw skills install sipeed/picoclaw-skills/weather" which fetches a GitHub repo such as https://github.com/sipeed/picoclaw-skills), and those fetched skill repositories are loaded into the workspace and can contain prompt templates or executable code that the agent will run—meeting the criteria for a runtime external dependency that can control prompts/execute code.