Partner reports that when using Atlassian-provided headless access paths (ACLI and MCP/Rovo MCP server) against Jira Cloud, they can only list attachments on issues via API, but there is no supported way to actually download the attachment files. Their explicit requirement is to allow agents/automations to retrieve the full binary content of Jira issue attachments, not just the metadata, so they can run downstream processing (AI analysis, automation, data pipelines, etc.) on what customers and partners upload.

They called out that this limitation severely blocks automation and analysis workflows because they cannot programmatically access the same information that is available in the Jira UI. Users are forced back to GUI-based/manual workflows instead of being able to use ACLI/MCP-driven integrations for attachments, which is particularly painful for AI use cases that depend on ingesting documents, logs, and other files attached to issues.

Today, Jira Cloud’s backend clearly supports downloading attachments via the REST APIs, but that capability is not surfaced through ACLI or MCP tools. From the customer’s perspective, this is experienced as “Jira via ACLI/MCP does not support downloading attachments” even though the base product does. The underlying gap on the Jira side is that there is no well-documented, tool-friendly way for these Atlassian-owned headless clients to retrieve attachment file content that is consistent with our security and permission model.