The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill appends user input via the $ARGUMENTS variable at the end of the SKILL.md file.
Ingestion points: The $ARGUMENTS placeholder at the end of the file directly interpolates user input into the agent's context.
Boundary markers: Absent. There are no delimiters or instructions to the agent to ignore or treat the appended text as untrusted data.
Capability inventory: The skill has the capability to execute shell commands (npm, git, curl, grep), which could be targeted if an attacker provides instructions that the agent interprets as its own.
Sanitization: Absent. User input is placed directly into the context.
[COMMAND_EXECUTION]: The skill executes several shell commands to perform code verification and deployment readiness checks.
Evidence: Use of npm test, npm run build, npm run lint, npx tsc, and grep for secret detection.
Context: These commands are standard for a development-focused skill and are used in accordance with its stated purpose of verifying code readiness.
[EXTERNAL_DOWNLOADS]: The skill includes placeholder commands for post-deployment health checks.
Evidence: curl -s https://your-app.com/health.
Context: This is a standard verification pattern using a placeholder domain for application monitoring.

ship