fix-flaky-test
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- Command Execution: The skill frequently uses system-level commands and CLI tools, including
dotnet,gh(GitHub CLI), anddocker. These are used to build projects, run tests, and manage the environment during the reproduction cycle. - Process and Environment Management: The provided reproduction scripts (
run-test-repeatedly.shandrun-test-repeatedly.ps1) perform environment cleanup by terminating specific processes (e.g.,dcp,dotnet-tests) and pruning Docker resources. These scripts include a manual confirmation prompt when run in a local environment to ensure the user is aware of the cleanup actions. - Data Ingestion from External Sources: The agent is instructed to gather data from GitHub issues and CI failure logs to identify root causes. This involves processing external text which is a standard debugging activity but constitutes an ingestion surface for untrusted data.
- Interaction with Trusted Repositories: All network operations and data retrievals are targeted at the official
microsoft/aspirerepository. These operations are used for downloading logs and pushing fix branches as part of the repository maintenance workflow.
Audit Metadata