The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a shell command template for the agent to verify backlink status (curl -sIL [URL] | grep -i rel=). This presents a command injection risk if the agent populates the URL from user input or external sources without strict sanitization of shell metacharacters.
[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its ingestion of external data.
Ingestion points: The instructions in SKILL.md direct the agent to read context from .agents/product-marketing-context.md before proceeding.
Boundary markers: No delimiters or instructions to ignore embedded commands are present in the instructions to protect the agent context from poisoned data.
Capability inventory: The skill uses curl via shell execution to perform network requests and data verification.
Sanitization: No validation, escaping, or filtering of the ingested context or the processed URLs is described in the skill scripts.
[EXTERNAL_DOWNLOADS]: The skill references a large volume of external directory URLs and software marketplaces across SKILL.md and references/directory-list.md. While most are well-known SEO and software review platforms, the high number of external links increases the attack surface for redirection to malicious or phishing domains.

directory-submissions