The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input via the [TICKET] and [description] arguments and uses them to construct shell commands and file content.
Ingestion points: User request arguments defined in argument-hint.
Boundary markers: None. The skill does not instruct the agent to sanitize or validate the format of the ticket key or description before use.
Capability inventory: Executes shell commands via npx playwright test (SKILL.md step 5) and performs file system writes to create spec files.
Sanitization: Absent. There is no logic to prevent characters like semicolons, backticks, or pipes in the ticket key from being interpreted by the shell.
[DYNAMIC_EXECUTION]: The skill generates TypeScript code based on a template (references/handover-spec-template.md) and then executes it using Playwright.
Evidence: The workflow involves creating a new .spec.ts file in e2e/tests/handover/ and immediately running it via npx playwright test.
[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command that incorporates user-provided variables.
Evidence: npx playwright test --project="chromium:handover" {TICKET}-{description}.spec.ts in SKILL.md. If a user provides a ticket name like TICKET; rm -rf /, this could lead to unintended command execution depending on how the agent platform handles shell interpolation.

create-handover-test