a1-yandex-kit-setup

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

• Potentially malicious external URL detected (high risk: 1.00). The helper intentionally runs "npx -y mcp-yandex-kit@latest" at runtime and probes/contacts registry.npmjs.org and api.kit.yandex.net (seen in scripts/smoke.mjs and SKILL.md), meaning the skill fetches and executes remote npm package code and relies on those external hosts.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I scanned the repository for literal high-entropy values that look like real API tokens. I flagged three string constants that are not placeholders and appear to be realistic tokens used in tests:
• scripts/modules.test.mjs:[16] const SECRET = "y0_AgAAAA-module-test";
• scripts/setup.test.mjs:[46] const SECRET_ONE = "y0_AgAAAA-secret-one";
• scripts/setup.test.mjs:[47] const SECRET_TWO = "y0_AgAAAA-secret-two";

These values are long, random-looking, and therefore meet the “high-entropy” criterion for secrets. I did not flag placeholders, environment variable names, or obvious simple example passwords (none of which were present as high-entropy literals here).

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

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The skill explicitly instructs the agent to modify user-level configuration files (run setup/configure, write tokens, create backups and set file modes) which changes the host machine's state and stores credentials, so it should be flagged.