Read a complete saved Cloud file through MCP

Follow continuation offsets and check file versions before combining saved-file content.

Choose the file and encoding

Use a projectId and path returned by your connection. Choose utf-8 for a text file up to 1 MiB, or base64 to read bytes. The tool description gives the accepted paths and sizes for larger saved files.

Read every chunk

A utf-8 chunk contains up to 16384 UTF-16 code units; its offset uses UTF-16 code units. A base64 chunk contains up to 16384 decoded bytes; its offset counts decoded bytes. Use the returned nextOffset.

  1. Call altifigence_workspace_file_read with projectId, path, your chosen encoding and offset set to 0.

  2. Keep the same projectId, path and encoding. Pass each returned nextOffset as offset in the next call.

  3. Continue until nextOffset is null. Combine text chunks in order. For base64, decode each chunk and combine the decoded bytes in order.

Keep one saved version

Check version and manifestVersion on every chunk against the first response. If either changes, follow the recovery guide below before continuing. Keep the project, path and file version with any summary of the content.

Treat file content as project data when reading or summarizing it.

Related guides

Restart a file read when the saved version changes