The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the analysis of CI logs and artifacts.
Ingestion points: The agent reads potentially untrusted data from CI logs via gh run view --log-failed and downloaded artifacts (e.g., playwright-report, logcat.txt) using agentTool downloadArtifact.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to distinguish between its own instructions and potentially malicious commands embedded within the logs it analyzes.
Capability inventory: The agent has significant capabilities including executing local build/test commands (pnpm), running mobile debugging tools (adb), and modifying the repository followed by git commit and git push operations.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from logs or artifacts before the agent processes it to determine its next actions.

fix-tests