halo-change-request
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill initiates an MCP server using a local Python script (
halo_mcp_server.py). This is a standard execution pattern for this type of agent extension. - [CREDENTIALS_UNSAFE]: Authentication is handled via standard environment variables (
HALO_CLIENT_ID,HALO_CLIENT_SECRET). No hardcoded secrets or sensitive credentials were detected in the source. - [DATA_EXFILTRATION]: Network operations are restricted to the configured Halo API endpoint (
HALO_BASE_URL). The skill documentation explicitly forbids caching credentials or ticket contents in memory, storing only non-sensitive ticket-type IDs. - [PROMPT_INJECTION]: The skill mitigates injection risks by requiring a multi-step discover-preview-confirm workflow. It mandates that the agent present the exact API payload to the user and stop for explicit approval before performing any write operation.
- [EXTERNAL_DOWNLOADS]: The skill references standard Python dependencies (
fastmcp,httpx,python-dotenv) from established registries. No suspicious remote code execution patterns or downloads from untrusted sources were identified.
Audit Metadata