The Agent Skills Directory

[COMMAND_EXECUTION]: The skill constructs a curl command for Slack notifications by interpolating variables like {{failed_details}} directly into a shell string. If test error messages contain shell metacharacters such as backticks or $(), they could be executed by the shell during the reporting process.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from test logs and JSON reports and incorporates it into reports and commands without sanitization. \n
Ingestion points: test-output.log and Playwright JSON reports. \n
Boundary markers: Absent. \n
Capability inventory: Shell command execution (curl, npx playwright) and file writing (test-reports/). \n
Sanitization: None; data is directly interpolated into templates.
[DATA_EXFILTRATION]: The skill transmits test result data to a non-whitelisted external destination specified by the SLACK_WEBHOOK_URL environment variable.
[EXTERNAL_DOWNLOADS]: The skill invokes npx playwright, which downloads and runs components of the Playwright testing framework. This is standard behavior for developer tools originating from a well-known source.

report