force-link-environment

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs fetching a host token and passing it as a CLI argument (--token ) to helper scripts, which requires the LLM to handle and potentially emit the secret verbatim (high-risk pattern).

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

• Third-party content exposure detected (high risk: 0.85). Phase 1.5 MCP Learn grounding fetches public Microsoft Learn HTML via mcp__plugin_power-pages_microsoft-learn__microsoft_docs_fetch, which is outsider-authored free text ingested into the agent context at runtime.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly calls mcp__plugin_power-pages_microsoft-learn__microsoft_docs_fetch at runtime to fetch and surface the Microsoft Learn page (https://learn.microsoft.com/en-us/power-platform/alm/custom-host-pipelines), and that fetched content is required and used to ground/alter prompts and decisions before proceeding.