The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute arbitrary shell commands for reproducing and fixing issues, including development server commands and project scripts (e.g., pnpm lint, kill-port, pnpm dev).
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It processes untrusted data from error logs, stack traces, and external web content (via Context7 or agent-browser) without explicit boundary markers or instructions to ignore embedded commands. This could allow malicious content to hijack the agent's file system or shell capabilities.
Ingestion points: Error logs, stack traces, external documentation, and browser snapshots.
Boundary markers: Absent.
Capability inventory: Shell access (Bash), file modification (Write, Edit), and browser automation (agent-browser).
Sanitization: None detected for interpolated data.
[DATA_EXFILTRATION]: The skill handles sensitive authentication data via agent-browser persistence features (e.g., auth.json, browser data profiles). While the instructions recommend excluding these files from version control, the management of active session tokens and browser profiles represents a potential data exposure surface.

troubleshoot