The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructions explicitly direct the agent to search for and read credentials from sensitive local files such as .env.local, .env.test, .env.example, and project documentation to facilitate authentication during testing procedures.
[PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection as it navigates, snapshots, and interprets content from arbitrary web pages which are considered untrusted sources.
Ingestion points: Web application UI content ingested via browser_snapshot and browser_navigate in SKILL.md.
Boundary markers: Absent. There are no instructions to use delimiters or to ignore embedded commands within the processed UI text.
Capability inventory: The skill has the ability to execute arbitrary SQL via plugin-supabase-supabase:execute_sql, perform file system discovery, and manipulate the browser environment.
Sanitization: None. The skill does not validate or sanitize data extracted from the browser before using it to drive subsequent logic or tool calls.
[COMMAND_EXECUTION]: The skill utilizes powerful tools for data verification and session management, including execute_sql via the Supabase plugin and browser_run_code_unsafe, which allows for code execution within the browser context. These tools are used for legitimate QA purposes but represent a high-privilege capability set.

test-qa