The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the alera CLI tool to manage Agent Canvas updates, revisions, and durable decisions within an Alera terminal session.
[PROMPT_INJECTION]: The skill provides explicit safety guardrails, instructing the agent to exclude arbitrary commands, shell text, and external requests from the canvas components it publishes.
[SAFE]: The skill processes untrusted data via alera canvas catalog and alera canvas events (ingestion points in SKILL.md), but mitigates indirect prompt injection risks by explicitly forbidding the interpretation of this data as commands and providing instructions to limit output to meaningful, non-executable UI changes (boundary markers and sanitization).

agent-canvas