Book full text cannot be fetched or delivered as a file #869

Open
opened 2026-08-16 16:06:34 +00:00 by coilyco-ops-gaming · 3 comments

A member asked for a short ancient text with its full contents uploaded as a file, and the request cannot be completed.

What exists: gutendex returns download and read URLs for public-domain books, exa returns search links and metadata only, and no tool returns a page's full text into context. The only write surface is a private, text-only scratchpad that is wiped on deployment, so a file written there is not a deliverable, and Discord access here is read-only.

Result: the member can be handed working links to the full text but not the text itself or a file containing it. This is a capability gap, filed per harness policy. No existing issue found covering it (searched: upload, full text, file).

A member asked for a short ancient text with its full contents uploaded as a file, and the request cannot be completed. What exists: gutendex returns download and read URLs for public-domain books, exa returns search links and metadata only, and no tool returns a page's full text into context. The only write surface is a private, text-only scratchpad that is wiped on deployment, so a file written there is not a deliverable, and Discord access here is read-only. Result: the member can be handed working links to the full text but not the text itself or a file containing it. This is a capability gap, filed per harness policy. No existing issue found covering it (searched: upload, full text, file).
Owner

Sounds like there's a few things here, but probably 1 of them is probably a beaver MCP fetch of anything on the gutendex domain. And playwright for that matter. Actually, playwright on all the websites for our MCP tools, especially their docs pages! Ideally also an openapi page listed explicitly if they have one. The beaver MCPs should have a resource informing the agent of the root url and docs page for that MCP

Sounds like there's a few things here, but probably 1 of them is probably a beaver MCP fetch of anything on the gutendex domain. And playwright for that matter. Actually, playwright on all the websites for our MCP tools, especially their docs pages! Ideally also an openapi page listed explicitly if they have one. The beaver MCPs should have a resource informing the agent of the root url and docs page for that MCP
Owner

Oh right sorry, I meant our custom web fetch tool on that domains

The tool iterates a ton of domains right now, need to move its configuration into the yaml config map instead of a list inside and env var

Oh right sorry, I meant our custom web fetch tool on that domains The tool iterates a ton of domains right now, need to move its configuration into the yaml config map instead of a list inside and env var
Owner

Epic labeled so we're encouraged to decompose and expand into of trying to 1 shot

Epic labeled so we're encouraged to decompose and expand into of trying to 1 shot
Sign in to join this conversation.
No milestone
No project
No assignees
2 participants
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
coilyco-gaming/sirens-echo#869
No description provided.