The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill involves auditing external web pages, which introduces a potential indirect prompt injection surface if an attacker-controlled website contains malicious instructions hidden in its HTML.
Ingestion points: Untrusted content enters the agent's context via lighthouse_audit, take_snapshot, and the axe CLI when processing external URLs.
Boundary markers: The skill does not provide instructions to delimit or ignore instructions found within the audited content.
Capability inventory: The agent is instructed to execute shell commands for testing and to modify the repository's source code based on the audit results.
Sanitization: No explicit sanitization or validation of the data retrieved from external sources is documented.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation and use of standard, well-known accessibility testing tools from the public NPM registry.
Evidence: Instructions include the use of npx lighthouse and npm install @axe-core/cli -g for automated auditing.

accessibility