configure-clarc

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

• Third-party content exposure detected (high risk: 0.85). The skill clones a public GitHub repository at runtime (git clone https://github.com/marvinrichter/clarc.git /tmp/clarc) and then copies/reads its outsider-authored SKILL.md and rule .md files into the installation target, which become LLM-readable context during subsequent use.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly runs "git clone https://github.com/marvinrichter/clarc.git" at runtime to fetch SKILL.md files (and noted scripts/hooks such as in continuous-learning-v2) which are then copied into the agent's skills/rules and thus directly control agent prompts and may include executable scripts—https://github.com/marvinrichter/clarc.git