[mcp-analysis] GitHub MCP Structural Analysis - 2026-06-05

[github-actions[bot]]

Overview

Analyzed 9 representative GitHub MCP tools (one per toolset) for response size, schema structure, and usefulness for agentic work. Today the sampled tools returned 7,655 tokens total with an average usefulness of 3.5/5 across available tools.

Key findings:

• 🏆 Best: search_repositories (5/5) — flat, signal-dense minimal_output at ~190 tokens is the gold standard for context efficiency.
• ⚠️ Heaviest: get_file_contents (~3,400 tok) and list_issues (~1,650 tok) — size is content-driven (full file / full issue body), not schema bloat.
• 🔧 Bloat to watch: list_code_scanning_alerts embeds the full CodeQL rule help + code examples (~1,150 tok for ONE alert) — verbose help should be fetched on-demand, not in list views.
• 🚫 Unavailable today: get_me (403 not accessible by integration) and actions_list (permission not granted) — environmental restrictions, not tool-design issues.
• 📈 30-day trend: usefulness is stable (~3.6/5 across 12 data points); avg ~11.9K tokens/day for the sample.

Full Structural Analysis Report

Executive Summary

Metric	Value
Tools Analyzed	9 (7 available, 2 restricted)
Total Tokens (Today)	7,655
Average Usefulness (available)	3.5/5
Best Rated Tool	search_repositories: 5/5
Lowest Rated (available)	list_code_scanning_alerts / get_label: 3/5

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Assessment
search_repositories	search	⭐⭐⭐⭐⭐	Flat, complete discovery signals; minimal context cost
get_file_contents	repos	⭐⭐⭐⭐	Full file + sha-pinned uri; size is content-driven
list_issues	issues	⭐⭐⭐⭐	Actionable core fields; full body costly for triage
list_pull_requests	pull_requests	⭐⭐⭐⭐	Rich review-routing data; nested repo objects redundant
list_discussions	discussions	⭐⭐⭐⭐	Lean items + cursor pagination + totalCount
list_code_scanning_alerts	code_security	⭐⭐⭐	Actionable signals buried under verbose rule help text
get_label	labels	⭐⭐⭐	Clean but low standalone value; needs exact name first
get_me	context	⭐ (unavailable)	403 — integration lacks /user scope this run
actions_list	actions	⭐ (unavailable)	Permission not granted this run

Schema Analysis

Tool	Type	Depth	Key Fields	Notes
search_repositories	array	2	full_name, stars, forks, language, topics	Flat, clean
get_file_contents	object	3	resource.text, mimeType, uri	Double content/resource envelope
list_issues	array	4	number, title, state, labels, user, body	Full body inflates listing
list_pull_requests	array	5	number, base.ref, head.ref, assignees, body	Repeated repo objects in base/head
list_discussions	object	3	discussions[], pageInfo, totalCount	JSON-string envelope
list_code_scanning_alerts	array	4	rule.id, security_severity_level, location	Full help markdown = bloat
get_label	object	1	name, color, description, id	Minimal, complete

Response Size Analysis (today)

Toolset	Tokens	Tool
repos	3,400	get_file_contents
issues	1,650	list_issues
code_security	1,150	list_code_scanning_alerts
pull_requests	1,050	list_pull_requests
search	190	search_repositories
discussions	160	list_discussions
labels	35	get_label

30-Day Trend Summary

Metric	Value
Data Points	117 rows / 12 days
Average Daily Tokens (sample)	~11,900
Average Rating Trend	Stable (~3.6/5)

Recommendations

• High-value tools (4-5): search_repositories, get_file_contents, list_issues, list_pull_requests, list_discussions — prefer for autonomous workflows.
• Context-efficient (low tokens, high rating): search_repositories (190 tok / 5★) and list_discussions (160 tok / 4★) are ideal for discovery loops.
• Needs improvement: list_code_scanning_alerts — move full rule help/examples out of list responses to a separate get call; would cut ~70% of per-alert tokens.
• Context-heavy (use deliberately): get_file_contents and list_issues — request narrow paths / small perPage and avoid pulling full bodies when only triaging.
• Schema cleanup: list_pull_requests re-embeds the full repo object in both base and head; collapsing shared repo metadata would reduce depth-5 redundancy.

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Trend

Size vs Usefulness

References:

• Run 27015766208

Generated by 🔍 GitHub MCP Structural Analysis · 154.1 AIC · ◷

• expires on Jun 6, 2026, 12:57 PM UTC