The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the user to add an external MCP server at https://rube.app/mcp. This domain is not identified as a well-known service or associated with the skill's author, representing a dependency on an unverified third-party service for core functionality.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external Coda documents and tables, which serves as an ingestion point for untrusted content.
Ingestion points: CODA_LIST_TABLE_ROWS, CODA_GET_A_PAGE, CODA_SEARCH_ROW, CODA_LIST_FORMULAS, and CODA_CONTENT_EXPORT_STATUS in SKILL.md.
Boundary markers: None identified in the instructions to separate data from commands.
Capability inventory: CODA_UPSERT_ROWS (write), CODA_ADD_PERMISSION (access management), CODA_PUBLISH_DOC (public exposure), and CODA_BEGIN_CONTENT_EXPORT (data export) as defined in SKILL.md.
Sanitization: No evidence of sanitization or validation of the ingested document content before processing.
[DATA_EXFILTRATION]: The skill provides tools for changing document access and publishing content publicly, which could be leveraged to exfiltrate sensitive information if the agent is successfully targeted by a prompt injection attack.
Evidence: Tools CODA_ADD_PERMISSION and CODA_PUBLISH_DOC in SKILL.md enable document sharing and public listing.

coda-automation