The Agent Skills Directory

[PROMPT_INJECTION] (LOW): Indirect Prompt Injection risk exists as the skill ingests content from Coda documents which could contain malicious instructions.
Ingestion points: Data is pulled into the agent context via CODA_LIST_TABLE_ROWS, CODA_GET_A_PAGE, CODA_SEARCH_ROW, and CODA_GET_A_FORMULA.
Boundary markers: The documentation does not specify the use of delimiters or 'ignore embedded instructions' warnings for processed data.
Capability inventory: The skill includes powerful tools such as CODA_UPSERT_ROWS (write access), CODA_ADD_PERMISSION (access control modification), and CODA_PUBLISH_DOC (public data exposure).
Sanitization: No explicit sanitization or validation steps are defined for the data retrieved from external Coda documents.
[EXTERNAL_DOWNLOADS] (LOW): The skill relies on an external MCP server endpoint (https://rube.app/mcp). While standard for the Rube/Composio ecosystem, it represents a dependency on a third-party service not listed in the trusted organizations list.
[PRIVILEGE_MANAGEMENT] (INFO): The skill provides tools like CODA_ADD_PERMISSION and CODA_PUBLISH_DOC which allow for significant changes to document security and visibility. These are intended features but require careful handling by the agent to prevent unauthorized access expansion.

coda-automation