The Agent Skills Directory

Indirect Prompt Injection Surface: The skill instructs the agent to read and analyze content from .zip trace files, which may include HTML snapshots, console logs, and network data. If a trace file contains malicious content, it could potentially influence the agent's reasoning or behavior during analysis.
Ingestion points: The skill reads external .zip files through commands like trace open and trace snapshot (SKILL.md).
Boundary markers: There are no explicit delimiters or instructions to the agent to disregard embedded content within the trace data.
Capability inventory: The agent can execute bash commands and perform network analysis via the Playwright CLI.
Sanitization: No explicit sanitization of the trace content is described before it is presented to the agent.
External Resource Fetching: The skill utilizes the npx command to run the playwright utility. This process involves retrieving and executing code from the public npm registry. The playwright package is maintained by a recognized organization, which is a common practice for this type of tool.
Dynamic Browser Scripting: The snapshot functionality allows for the execution of JavaScript via the eval flag within a headless browser environment. This is used to query the state of a page snapshot and should be monitored for the complexity of scripts being executed.

playwright-trace