microsoft / microsoft/mcp

[Feature] docs_item-definitions / docs_platform-api-spec responses routinely exceed token limits with no filter option

Open
#3,254 1 comment 0 reactions 2 assignees Claimed by @AmosHersch View on GitHub
customer-reported enhancement needs-triage server-Fabric.Mcp
Dominant language
C#
Stars
3.7k
Forks
624
Avg merge
2d 20h
Merged PRs (30d)
220

Description

**Describe the bug**
This is a feature request. `docs_item-definitions` and `docs_platform-api-spec` responses routinely exceed the calling client's token/size limits (observed: 83K-715K characters for a single call) and get written to a temp file instead of being returned inline. This works, but turns every schema lookup into a two-step "call tool, then grep/read the dumped file for the one section you need" process.

**Expected behavior**
An optional filter/search parameter (e.g. `section`, `activity-type`, or a free-text `grep`) on `docs_item-definitions` that returns only the matching subsection of the schema, so a targeted lookup (e.g. "what are TridentNotebook's required properties") doesn't require downloading and searching the entire `dataPipeline` schema document.

**Actual behavior**
Example: looking up the `TridentNotebook` activity schema within the `dataPipeline` item-definition required:
1. Calling `docs_item-definitions(workload-type="dataPipeline")`
2. Receiving a "result exceeds maximum allowed tokens" error with the full content saved to a file
3. Manually grepping that file for "notebook" to find the relevant subsection

This pattern repeated for `docs_platform-api-spec` as well (715K characters in one response).

**Reproduction Steps**
1. Call `docs_item-definitions(workload-type="dataPipeline")`
2. Observe the response is too large to return inline and gets redirected to a file
3. Note there is no parameter to narrow the request to a specific activity type or section

**Environment**
- Fabric MCP Server
- Windows 11

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.