playwright-best-practices-for-agents
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill guides the agent to read and process accessibility snapshots and error context files derived from external websites. This creates a surface where untrusted content from a webpage could potentially influence the agent's logic.\n
- Ingestion points:
playwright-cli snapshotoutput anderror-context.mdfiles referenced in SKILL.md and debugging.md.\n - Boundary markers: The instructions lack explicit directives for the agent to treat data from the accessibility tree as untrusted or to ignore embedded instructions.\n
- Capability inventory: The agent has access to shell execution via
npx playwrightandplaywright-cli, and can perform file system operations.\n - Sanitization: There is no mention of filtering or sanitizing the snapshots to remove potentially malicious instruction patterns before processing.\n- [EXTERNAL_DOWNLOADS]: The skill recommends the installation of established packages from the NPM registry, including
@playwright/test,@playwright/cli,otpauth, anddotenv. These are well-known tools from recognized sources in the developer community.
Audit Metadata