update-claude-code

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

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow (SKILL.md step 4) explicitly fetches the public CHANGELOG from GitHub (https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md) using gh api or curl and requires the agent to read and summarize those third-party contents, which could influence its outputs/decisions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly fetches the changelog at runtime (e.g., via curl https://raw.githubusercontent.com/anthropics/claude-code/main/CHANGELOG.md and the GitHub URL https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md) and requires that external content to generate the agent's update summary, so external content directly controls the agent's output.