The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses sensitive environment configuration files, including .env, .env.local, .env.test, and .envrc, as seen in e2e-test-execution.md. It searches these files for patterns such as TEST_USER, ADMIN_PASSWORD, and TEST_EMAIL to automate authentication flows during browser testing. While intended for testing purposes, this behavior involves accessing files that often contain production or development secrets.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted content from external sources. It uses the GitHub CLI to fetch pull request bodies and associated issue descriptions in e2e-test-execution.md to guide its visual verification logic. This untrusted data is processed without explicit boundary markers or sanitization, potentially allowing an attacker to influence the agent's behavior via malicious instructions embedded in a GitHub issue or PR.
Ingestion points: gh pr view and gh issue view commands located in e2e-test-execution.md.
Boundary markers: None identified.
Capability inventory: The skill has access to the Bash tool for system command execution, the Edit and Write tools for modifying the local file system, and browser-based script evaluation tools.
Sanitization: None identified.
[COMMAND_EXECUTION]: The skill frequently utilizes the Bash tool to execute system-level operations, such as git rebasing, running build tools (go build, npm), and managing database migrations. Additionally, it uses mcp__chrome-devtools-mcp__evaluate_script to inject and execute JavaScript routines within the browser for page stabilization and verification purposes.

e2e-verify