kuroco-docs

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). This is a direct raw GitHub link to a shell script that the skill tells users to curl and pipe to bash — executing remote .sh files without inspecting the content is a high-risk practice even if hosted on GitHub, so treat it as suspicious unless you fully trust and have reviewed the repository and script.

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

• Third-party content exposure detected (high risk: 1.00). SKILL.md explicitly instructs syncing external documentation (via curl of the GitHub script and scripts/sync-docs.sh), and that script downloads and unpacks a ZIP from https://rcms.g.kuroco-img.app/files/user/skills/current.zip into the docs/ directory which the agent is expected to read/search and therefore could contain untrusted third‑party instructions influencing its behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs at runtime to run remote code via curl -fsSL https://raw.githubusercontent.com/diverta/kuroco-skills/main/skills/kuroco-docs/scripts/sync-docs.sh | bash (and that script in turn downloads https://rcms.g.kuroco-img.app/files/user/skills/current.zip to populate docs the skill relies on), so external code/content fetched at runtime is executed and directly controls the agent's document content and behavior.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (medium risk: 0.60). The skill tells the agent to run a remote script via "curl ... | bash" to sync and modify files under the plugin root—this alters the host filesystem (though it does not explicitly request sudo, create users, or edit system config files), so it poses a moderate risk of compromising machine state.