[mcp-analysis] MCP Structural Analysis - 2026-03-31

[github-actions[bot]]

Analysis of 9 GitHub MCP tool responses across 8 toolsets for 2026-03-31. Average usefulness rating: 3.56/5. High-value tools (get_file_contents, search_repositories) deliver complete, actionable data efficiently. Context bloat risk remains in code_security (up to 34K+ tokens when alerts exist) and labels (444 labels returned without pagination). This is the second daily run — 2-day trend data now available.

Key Findings

Metric	Value
Tools Analyzed	9
Total Tokens Today	9,875
Avg Usefulness Rating	3.56/5
Best Tool	get_file_contents — 5/5
Worst Tool	get_me — 1/5 (403 error)
Context Risk Tools	code_security, labels, actions

Usefulness by Toolset

Rating	Toolset	Tool	Assessment
⭐⭐⭐⭐⭐	repos	get_file_contents	Complete, SHA-addressable, immediately actionable
⭐⭐⭐⭐⭐	search	search_repositories	Compact, clean metadata; best context efficiency
⭐⭐⭐⭐	issues	list_issues	Good data; body size varies widely
⭐⭐⭐⭐	pull_requests	list_pull_requests	Comprehensive; draft flag useful
⭐⭐⭐⭐	discussions	list_discussions	Most compact list tool; perPage respected
⭐⭐⭐	actions	list_workflows	perPage ignored; 5 URL fields per entry = bloat
⭐⭐⭐	code_security	list_code_scanning_alerts	No pagination; 0 today but 138K chars yesterday
⭐⭐⭐	labels	list_label	444 labels, no pagination; clean schema
⭐	context	get_me	403 error in this integration

Full Structural Analysis Report

Schema Analysis

Tool	Data Type	Nesting Depth	Key Fields	Pagination
get_me	error	1	—	N/A
list_issues	array	3	number, title, state, body, labels[], user, timestamps	perPage ✅
list_pull_requests	array	4	number, title, state, body, base{}, head{}, user, labels[], draft, merged	perPage ✅
list_workflows	object	3	id, name, path, state, created_at, updated_at, url×5	perPage ❌ ignored
get_file_contents	object	2	sha, mimeType, text	N/A
list_discussions	object	3	number, title, category{}, user{}, html_url, timestamps	perPage ✅, cursor ✅
list_code_scanning_alerts	array	4	number, rule{}, severity, state, location{}, tool{}	❌ returns all
list_label	object	2	id, name, color, description	❌ returns all 444
search_repositories	array	2	name, full_name, description, language, topics[], stats	perPage ✅

Response Size Analysis

Toolset	Tokens (Today)	Tokens (Yesterday)	Delta	Notes
repos	3,500	2,100	+67%	README size varies
code_security	5	34,584	-99.99%	Highly volatile — depends on open alerts
actions	2,000	2,025	~stable	perPage ignored; ~30 entries returned
labels	2,000	1,850	+8%	Stable; returns all labels
issues	1,400	850	+65%	Body size drives variance
pull_requests	600	550	+9%	Stable
search	200	350	-43%	Efficient; minimal_output=true
discussions	155	155	stable	Most consistent tool
context	15	55	—	Error response

Tool-by-Tool Analysis

Tool	Toolset	Tokens	Schema	Depth	Rating	Notes
get_me	context	15	error	1	⭐	403 error; integration limitation
list_issues	issues	1,400	array	3	⭐⭐⭐⭐	Body inflation risk; labels embedded
list_pull_requests	pull_requests	600	array	4	⭐⭐⭐⭐	base/head repos add depth; draft flag valuable
list_workflows	actions	2,000	object	3	⭐⭐⭐	5 URL variants per workflow; per_page ignored
get_file_contents	repos	3,500	object	2	⭐⭐⭐⭐⭐	SHA enables caching; file size drives tokens
list_discussions	discussions	155	object	3	⭐⭐⭐⭐	Best pagination; cursor + totalCount present
list_code_scanning_alerts	code_security	5	array	1	⭐⭐⭐	Empty today; can explode to 34K+ without pagination
list_label	labels	2,000	object	2	⭐⭐⭐	All 444 labels returned; clean flat schema
search_repositories	search	200	array	2	⭐⭐⭐⭐⭐	Most context-efficient; minimal_output default

30-Day Trend Summary (2 data points)

Metric	2026-03-30	2026-03-31	Trend
Total Daily Tokens	40,469	9,875	↓ (code_security was 34K yesterday)
Avg Usefulness Rating	3.44/5	3.56/5	↑ Stable/improving
Tools Analyzed	9	9	Stable
Context Risk Events	1 (code_security)	0	↓

Recommendations

High-value tools (rating 4–5, prefer these):

• search_repositories — best tokens-to-value ratio
• list_discussions — compact, well-paginated
• list_pull_requests — good structure, use perPage

Tools needing improvement:

• list_code_scanning_alerts — needs pagination support urgently; unpredictable context cost
• list_label — should support pagination; 444 labels is heavy for large repos
• list_workflows — perPage parameter is silently ignored

Context-efficient tools (< 500 tokens, rating ≥ 4):

• search_repositories (200 tokens, ⭐⭐⭐⭐⭐)
• list_discussions (155 tokens, ⭐⭐⭐⭐)

Context-heavy tools (> 2,000 tokens) — use with caution:

• get_file_contents (varies; large files = large context)
• code_security (0 to 34K+ tokens depending on repo state)

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Trend

Size vs Usefulness

References:

• §23794579594

AI generated by GitHub MCP Structural Analysis · history

• expires on Apr 1, 2026, 11:36 AM UTC

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-04-01T11:36:30.792Z.

Closed by Workflow