file-test-bug

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • Local File Access and Information Transfer: The skill reads local test reports (junit.xml), execution metadata (agent-metadata.md), and test source files from the local filesystem. This data is then used to populate a GitHub issue in the microsoft/github-copilot-for-azure repository. While this involves the transmission of local data to an external repository, the activity is a direct component of the skill's primary purpose for bug reporting within a development context.\n- Indirect Prompt Injection Surface: By analyzing the contents of agent-metadata.md to diagnose test failures, the agent processes data that may be generated by other processes. If these logs were to contain content intended to influence agent behavior, it could potentially affect the diagnosis or the body of the generated GitHub issue.\n
  • Ingestion points: The skill ingests data from tests/reports/junit.xml and agent-metadata.md as specified in the steps within SKILL.md.\n
  • Boundary markers: No explicit delimiters or instructions to disregard embedded commands in the log files were identified.\n
  • Capability inventory: The skill has the capability to write data to a remote repository using the github-mcp-server-create_issue tool.\n
  • Sanitization: The skill does not perform programmatic sanitization or filtering of the log content before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 12:48 AM