Reported issues for zenodo-mcp-server
Pod holds 7 of 7 GitHub reports that passed its relevance review. This can include external user reports, maintainer-confirmed bugs, and concrete feature gaps. Treat them as evidence to inspect, not a count of distinct defects.
Back to zenodo-mcp-server.
Most discussed
bug(zenodo_list_files): archive-mode notices cite a missing download_url and hide readable unlisted members
Server version
0.1.0
mcp-ts-core version
0.13.6
Runtime
Node
Runtime version
Node 26.5.0
Transport
HTTP
Description
In archive mode (archive_key set), two zenodo_list_files notices send the agent somewhere the response doesn't support:
- When Zenodo truncates the listing, the notice says to "download the archive from
download_url", but the archive-mode output has nodownload_url. Thearchiveobject carries onlykey,size,listed_members,…
Read the thread · 2026-09-24 · closed · 3 comments
bug(zenodo_read_file): ZIP-member truncation notice says to download the archive when a larger max_bytes would do
Server version
0.1.0
mcp-ts-core version
0.13.6
Runtime
Node
Runtime version
Node 26.5.0
Transport
HTTP
Description
When a zenodo_read_file ZIP-member read stops at max_bytes, the notice always says to download the whole archive, even when the member is smaller than the 65,536-byte cap. Calling again with a larger max_bytes would return the full member, but the notice doesn't say that.
Steps to reproduce
zenodo_read_file`{ "id": "22705923",…
Read the thread · 2026-09-24 · closed · 3 comments
bug(zenodo_search_records): resource_type rejects the type::id search_value lookup returns
Server version
0.1.0
mcp-ts-core version
0.13.6
Runtime
Node
Runtime version
Node 26.5.0
Transport
HTTP
Description
zenodo_lookup_vocabulary (resource_types) prints each entry's search_value in <type>::<id> form, which is Zenodo's own spelling of a subtype. zenodo_search_records rejects that exact value as resource_type, with the message Missing required field, which reads as if the parameter were absent. publication::publication-article maps to…
Read the thread · 2026-09-24 · closed · 3 comments
feat(zenodo_read_file): sniff extensionless octet-stream files instead of refusing them
Use case
zenodo_read_file decides from the MIME type and extension alone whether a file is text, and it does so before fetching anything. Files like LICENSE, COPYING, Makefile, Dockerfile, and README arrive from Zenodo as application/octet-stream with no extension, so they come back not_text with no content. They are among the files an agent most often needs to read in a software deposit.
Proposed behavior
For application/octet-stream (or no MIME type) with an…
Read the thread · 2026-09-24 · closed · 2 comments
bug(identifiers): wrapped id with trailing punctuation is rejected as invalid_identifier
Server version
0.1.0
mcp-ts-core version
0.13.6
Runtime
Node
Runtime version
Node 26.5.0
Transport
HTTP
Description
parseRecordRef strips wrapping <…> before it removes trailing punctuation. An id that has both, such as a DOI copied out of a sentence like …available at <https://doi.org/…>., is rejected as invalid_identifier. Each form works on its own.
Steps to reproduce
zenodo_get_record with each id:
| id | result |
|---|---|
| … |
Read the thread · 2026-09-24 · closed · 2 comments
bug(zenodo_search_records): facet counts ignore resource_type, file_type, and access_status filters
Server version
0.1.0
mcp-ts-core version
0.13.6
Runtime
Node
Runtime version
Node 26.5.0
Transport
HTTP
Description
The zenodo_search_records description and facets schema say facet counts cover "the full match set", but resource_type, file_type, and access_status do not narrow them. Zenodo applies those three as post-filters, so the aggregations are computed before they apply. The counts then contradict total, and an agent reading `file_type: csv…
Read the thread · 2026-09-24 · closed · 2 comments
bug(zenodo_list_versions): page past the 10,000 window returns a raw upstream 400
Server version
0.1.0
mcp-ts-core version
0.13.6
Runtime
Node
Runtime version
Node 26.5.0
Transport
HTTP
Description
zenodo_list_versions sends any page straight upstream. When page × size exceeds 10,000, Zenodo answers HTTP 400 and the tool surfaces the raw upstream error with no reason and no recovery. zenodo_search_records and zenodo_lookup_vocabulary both pre-validate the same window and return a typed result_window_exceeded.
Steps to…
Read the thread · 2026-09-24 · closed · 2 comments
Most recent
The remaining reports are on the project's issue tracker.