[mcp-analysis] MCP Structural Analysis - 2026-02-24 #18117
Closed
Replies: 1 comment
-
|
This discussion was automatically closed because it expired on 2026-02-25T11:23:31.660Z.
|
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
11 GitHub MCP tools analyzed across 3 days (Feb 22–24, 2026). Today's total context footprint: 23,926 tokens across 11 tools. Average usefulness rating: 3.09/5. Two tools remain inaccessible (403), while
list_discussionsandsearch_repositorieslead with perfect 5/5 ratings.list_code_scanning_alertscontinues to dominate token consumption at 13,514 tokens per call.Quick Summary
list_discussions,search_repositories— 5/5get_me,list_notifications— 1/5 (403 errors)list_code_scanning_alerts— 13,514 tokensget_label— 25 tokensFull Structural Analysis Report
Usefulness Ratings for Agentic Work
list_discussionssearch_repositorieslist_commitslist_issuesget_labellist_pull_requestslist_workflowsper_page=1(returns all 205)list_code_scanning_alertsget_file_contentsget_melist_notificationsSchema Analysis
get_meget_file_contentslist_commitslist_issueslist_pull_requestslist_workflowslist_code_scanning_alertslist_discussionsget_labelsearch_repositorieslist_notificationsResponse Size Analysis
list_code_scanning_alertslist_pull_requestslist_workflowslist_issuesget_label(+list_labelFeb 22)search_repositoriesget_file_contents,list_commitslist_discussionsget_melist_notifications*
get_label(single item). ** Feb 22 includedlist_label(2,755 tokens) which inflates this average.Tool-by-Tool Analysis — Feb 24, 2026
get_meget_file_contentslist_commitslist_issueslist_pull_requestslist_workflowslist_code_scanning_alertslist_discussionsget_labelsearch_repositorieslist_notifications3-Day Trend Summary
The Feb 22 spike was driven by
list_code_scanning_alertsreturning 25,476 tokens (vs ~13,514 on subsequent days, reflecting fewer open alerts in the payload), plus the inclusion oflist_label(2,755 tokens). Token usage has stabilized around ~24k/day for the current 11-tool test set.Recommendations
High-value tools (rating 4–5, use freely):
list_discussions(105 tokens),search_repositories(135 tokens),list_commits(225 tokens),get_label(25 tokens),list_issues(variable, ~600–1400 tokens)Context-efficient sweet spot (low tokens + high rating):
get_label→ 25 tokens / 4⭐list_discussions→ 105 tokens / 5⭐search_repositories→ 135 tokens / 5⭐list_commits→ 225 tokens / 4⭐Tools needing improvement:
list_code_scanning_alerts— Needs per-page filtering; currently returns entire alert set regardless of paramslist_workflows/list_pull_requests—per_pageparameter ignored or not enforced; payload redirect adds frictionget_file_contents— Should return file content directly rather than saving to disk silentlyUnavailable in this environment (403):
get_me,list_notifications— Integration token does not have user-level scopesVisualizations
Response Size by Toolset (3-Day Comparison)
Usefulness Ratings by Toolset
Daily Token Usage Trend
Token Size vs Usefulness (Scatter)
References: §22348447807
Beta Was this translation helpful? Give feedback.
All reactions