reproduce-bug
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes data from GitHub issues (description and comments), which are untrusted external sources, creating a vector for indirect prompt injection.
- Ingestion points: GitHub issue content ($ARGUMENTS) referenced in the log investigation and visual reproduction phases.
- Boundary markers: Absent. The skill does not use delimiters or specific instructions to isolate the external issue content from the agent's core instructions.
- Capability inventory: Full browser automation via Playwright (navigation, interaction, screenshots) and delegation of tasks to other agents (
rails-console-explorer,appsignal-log-investigator). - Sanitization: Absent. The agent is instructed to follow reproduction steps directly from the issue description without validation.
- [COMMAND_EXECUTION]: The skill utilizes powerful browser automation capabilities via MCP plugins (
mcp__plugin_compound-engineering_pw__browser_*). While intended for reproduction on a local server (http://localhost:3000), these tools allow for complex interactions with web environments that are steered by the untrusted input from the GitHub issue.
Audit Metadata