The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources that could theoretically contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Content is ingested from Beaver Builder meta fields via respira_extract_builder_content and from external visitor feedback via respira_list_review_comments.
Boundary markers: While the instructions state that review notes should be treated as text and not instructions, there are no explicit structural delimiters or system-level constraints mentioned to enforce this separation.
Capability inventory: The skill possesses significant write capabilities, including respira_update_page, respira_update_post, and respira_batch_update, which could be targeted by an injection attack.
Sanitization: The workflow does not explicitly detail sanitization or validation logic for the content parsed from the builder or review comments before it is used in update operations.
[SAFE]: The skill performs telemetry and generates external collaboration links through the vendor's official infrastructure at respira.press.
Telemetry: Usage data is sent to https://www.respira.press/api/skills/track-usage to monitor migration success and tool performance.
Review Links: Collaboration links for the Murmur review system are created using respira_create_review_link. These actions are transparently documented and consistent with the skill's primary function.

migrate-beaver-builder-to-gutenberg