The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous read-only shell commands using curl, grep, and jq for the agent to verify that the target website correctly serves protocol discovery files and metadata (e.g., agent-card.json, mcp.json, openapi.json).- [INDIRECT_PROMPT_INJECTION]: The skill processes content from external websites (using URLs provided in $ARGUMENTS), creating a surface where malicious content on a target site could attempt to influence the agent's behavior during the implementation or verification process.
Ingestion points: Website HTML and JSON metadata files fetched via curl commands in SKILL.md and reference guides.
Boundary markers: Absent; the agent is instructed to directly evaluate the output of the verification commands without specific safety delimiters.
Capability inventory: Access to curl, grep, and jq for network verification; implied file writing capabilities to implement protocol documents.
Sanitization: No explicit sanitization or validation of the fetched external data is performed before it enters the agent's context.

agent-protocols