The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it processes untrusted data from external sources to define the scope and execution of its testing operations.\n
Ingestion points: The skill instructions in SKILL.md direct the agent to analyze branch diffs, PR descriptions, and tickets to determine the UAT scope and acceptance slices.\n
Boundary markers: Absent. The instructions do not specify the use of delimiters or 'ignore' directives when processing these external inputs.\n
Capability inventory: The skill has extensive capabilities including arbitrary shell command execution, file system writes, and network requests via curl (referenced in references/cli-api-sdk-playbook.md).\n
Sanitization: Absent. There is no requirement for validation or escaping of the ingested data before it is used to influence the agent's testing plan.\n- [EXTERNAL_DOWNLOADS]: The skill utilizes dynamic installation patterns to fetch necessary testing skills and tools from the NPM registry.\n
Evidence: SKILL.md and references/web-electron-playbook.md include instructions to install tools such as agent-browser, playwright, chrome-cdp, electron, and pp-agent-capture if they are not already available. These are standard tools for application and browser automation.\n- [COMMAND_EXECUTION]: The skill's primary function involves executing various shell commands and scripts to perform acceptance testing, including building projects and interacting with local services.\n
Evidence: The playbooks in references/ provide templates for running commands like npm run dev, curl, and node to exercise the application under test and capture output using tee and jq.

user-acceptance