[mcp-analysis] MCP Structural Analysis - 2026-05-07

[github-actions[bot]]

Today's run analyzed 9 GitHub MCP tools across 9 toolsets. The discussions toolset leads with a perfect 5/5 usefulness score and only 113 tokens, while code_security and labels are the biggest context hogs due to embedded full documentation text and unbounded list returns. Two toolsets (context, search) were unavailable due to auth/rate-limit errors.

Full Structural Analysis Report

Executive Summary

Metric	Value
Tools Analyzed	9
Total Tokens (Today)	12,464
Average Usefulness Rating	2.8 / 5
Best Rated Tool	list_discussions — 5/5
Worst Rated Tools	get_me, search_code — 1/5 (unavailable)
Most Context-Expensive	get_file_contents — 3,750 tokens
Most Context-Efficient	list_discussions — 113 tokens

---

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Assessment
list_discussions	discussions	⭐⭐⭐⭐⭐	Most compact and well-structured response tested
list_issues	issues	⭐⭐⭐⭐	Good metadata; body text can balloon token count
list_pull_requests	pull_requests	⭐⭐⭐⭐	Rich data with nested refs; same body inflation risk
get_file_contents	repos	⭐⭐⭐	Full file text returned; expensive on large files
list_workflows	actions	⭐⭐⭐	Good fields; per_page parameter ignored (30 returned)
list_code_scanning_alerts	code_security	⭐⭐	rule.help embeds full CWE documentation (~3,500 chars/alert)
list_label	labels	⭐⭐	Returns all 517 labels; perPage not honored
get_me	context	⭐	403 Forbidden — integration token lacks user read scope
search_code	search	⭐	429 Rate Limited — unreliable for agentic automation

---

Schema Analysis

Tool	Type	Depth	Key Fields	Notes
get_me	error	1	error	Auth failure
list_issues	array	3	number, title, state, body, labels[], user{}	labels is array of objects
list_pull_requests	array	3	number, title, state, body, base{}, head{}, user{}	base/head each embed repo
get_file_contents	object	3	content[], mimeType, text, uri, sha	text is raw file content
list_workflows	object	2	total_count, workflows[]: id, name, path, state	pagination not respected
list_code_scanning_alerts	array	4	number, rule{id, help, severity}, tool{}, most_recent_instance{}	help is massive embedded doc
list_discussions	object	2	discussions[], pageInfo{cursor}, totalCount	cleanest pagination model
list_label	object	2	labels[]: id, name, color, description, totalCount	flat but unbounded
search_code	error	1	error	Rate limit

---

Response Size Analysis

Toolset	Tokens	Tool Tested
repos	3,750	get_file_contents
actions	2,875	list_workflows
code_security	2,000	list_code_scanning_alerts
labels	1,750	list_label
issues	1,000	list_issues
pull_requests	950	list_pull_requests
discussions	113	list_discussions
context	13	get_me (error)
search	13	search_code (error)

---

30-Day Trend Summary

Metric	Value
Data Points	9 (first run — baseline established)
Total Tokens Today	12,464
Average Daily Tokens	12,464
Trend	Baseline — subsequent runs will establish trends

---

Recommendations

High-value tools (rating 4–5):

• list_discussions — prefer for listing/pagination tasks; most efficient
• list_issues and list_pull_requests — reliable, but filter by state and avoid fetching when body content is not needed

Tools needing structural improvements:

• list_code_scanning_alerts — the rule.help field (full CWE/OWASP documentation) should be opt-in, not default
• list_label — must honor perPage to avoid returning all 517 labels
• list_workflows — per_page parameter not respected; always returns default page size

Context-efficient tools (low tokens + high rating):

• list_discussions — 113 tokens, rating 5

Context-heavy tools to use sparingly:

• get_file_contents — 3,750 tokens for README alone; avoid on large files without knowing file size first
• list_workflows — 2,875 tokens; consider caching workflow IDs after first fetch

Unavailable toolsets requiring attention:

• context (get_me) — 403 suggests the integration token needs read:user scope for user context
• search (search_code) — 429 rate limits make it unsuitable for repeated agentic calls; use list_* tools + local filtering instead

---

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Total

Token Size vs Usefulness

References: §25495597148

Generated by GitHub MCP Structural Analysis · ● 409.3K · ◷

• expires on May 8, 2026, 12:40 PM UTC

[github-actions[bot]]

This discussion has been marked as outdated by GitHub MCP Structural Analysis.

A newer discussion is available at Discussion #31012.