The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a process for auditing external pricing pages, which requires the agent to ingest and analyze untrusted data from URLs or pasted text. This creates a surface for indirect prompt injection if the source content contains malicious instructions designed to influence the agent's behavior.
Ingestion points: The SKILL.md instructions and references/pricing-page-teardown.md guide the agent to perform teardowns on live pricing pages by fetching their content or using the 'paste test' with rendered text.
Boundary markers: The instructions do not specify any delimiters or safety prompts to isolate the external pricing page data from the agent's core instructions.
Capability inventory: The skill is strictly instructional and does not provide any scripts, executables, or tools. It lacks capabilities for file system modification, shell command execution, or custom network requests, relying instead on the platform's native browsing tools.
Sanitization: No sanitization or validation protocols are established for the external data ingested during the teardown process.
[NO_CODE]: The skill package is comprised solely of Markdown documentation and evaluation configurations. The absence of executable code (scripts, binaries, or active configuration) effectively eliminates technical attack vectors such as remote code execution, privilege escalation, and persistence.

pricing