[mcp-analysis] MCP Structural Analysis - 2026-08-20

[github-actions[bot]]

Overview

Analyzed 10 representative GitHub MCP tools across all major toolsets on 2026-08-20. Average usefulness rating was 3.4/5, up slightly from 3.1/5 on 2026-08-19. The three mcpscripts wrappers (actions, labels) and the injected workflow context block scored a perfect 5/5 — flat JSON with no fixed overhead. The weakest performers were get_teams (blocked by permissions, 2nd day in a row) and list_issues (silent integrity-policy redaction with no explanation).

Metric	Value
Tools Analyzed	10
Total Tokens (Today)	2,695
Average Usefulness Rating	3.4/5
Best Rated Tools	github_context, list_label_wrapper, list_workflows_wrapper: 5/5
Worst Rated Tool	get_teams: 1/5 (blocked)

Full Structural Analysis Report

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Assessment
github_context	workflow_context	⭐⭐⭐⭐⭐	Excellent — zero-cost, flat, immediately actionable identity facts
list_label_wrapper	labels	⭐⭐⭐⭐⭐	Excellent — flat object, no _meta overhead
list_workflows_wrapper	actions	⭐⭐⭐⭐⭐	Excellent — clean JSON, perPage honored correctly
list_pull_requests	pull_requests	⭐⭐⭐⭐	Good — compact, field-filtered, actionable refs/SHAs
search_code	search	⭐⭐⭐	Adequate — useful total_count, but icon overhead dominates
get_file_contents	repos	⭐⭐⭐	Adequate — fixed ~2.2KB icon overhead regardless of file size
list_discussions	discussions	⭐⭐⭐	Adequate — clean cursor pagination, same icon overhead
list_code_scanning_alerts	code_security	⭐⭐⭐	Adequate — empty result today; severity param only accepts one enum value, not a list
list_issues	issues	⭐⭐	Limited — silent integrity-policy redaction, no explanation surfaced
get_teams	context	⭐	Poor — blocked by permission grant, 2nd consecutive day

Schema Analysis

Tool	Type	Depth	Key Fields	Notes
github_context	markdown	1	actor, repository, workspace, workflow-run-id	Prompt-injected, no MCP call
list_label_wrapper	object	2	item_count, labels[].name/color/description/id	No _meta bloat
list_workflows_wrapper	object	2	total_count, workflows[].name/path/state/html_url	No _meta bloat
list_pull_requests	array	3	number, title, state, head.ref/sha, base.ref/sha	Field-filtered
search_code	object	2	items[].name/path, total_count, incomplete_results	_meta icon overhead
get_file_contents	object	3	content[].resource.text/uri, _meta.serverInfo	Fixed ~2.2KB overhead independent of file size
list_discussions	object	3	discussions[].category/title/user, pageInfo, totalCount	_meta icon overhead
list_code_scanning_alerts	array	0	(empty this run)	Single-severity enum only
list_issues	array	1	number, title, state, labels, comments	Redacted result
get_teams	n/a	0	(blocked)	Permission not granted

Response Size Analysis

Toolset	Avg Tokens	Tools Tested
repos	871	1
discussions	696	1
search	626	1
actions	172	1
labels	106	1
pull_requests	104	1
workflow_context	75	1
issues	40	1
code_security	5	1
context	0	1 (blocked)

Tool-by-Tool Analysis

Tool	Toolset	Tokens	Schema	Rating	Notes
get_file_contents	repos	871	object/3	⭐⭐⭐	Fixed icon overhead worse ratio on small file than yesterday's larger README
list_discussions	discussions	696	object/3	⭐⭐⭐	totalCount=9800 useful; icon overhead constant
search_code	search	626	object/2	⭐⭐⭐	fields filtering kept items minimal
list_workflows_wrapper	actions	172	object/2	⭐⭐⭐⭐⭐	Best size-to-value ratio, consistent both days
list_label_wrapper	labels	106	object/2	⭐⭐⭐⭐⭐	Far leaner than native get_label (375 tokens yesterday)
list_pull_requests	pull_requests	104	array/3	⭐⭐⭐⭐	Consistent with yesterday
github_context	workflow_context	75	markdown/1	⭐⭐⭐⭐⭐	Zero MCP cost
list_issues	issues	40	array/1	⭐⭐	Silent redaction again, 2nd day
list_code_scanning_alerts	code_security	5	array/0	⭐⭐⭐	Empty for severity=critical; enum doesn't support comma lists
get_teams	context	0	n/a	⭐	Permission blocked, 2nd day

30-Day Trend Summary

Metric	Value
Data Points	20 (2 days)
Average Daily Tokens	4,936
Average Rating Trend	Improving (3.1 → 3.4)

Recommendations

• High-value tools (rating 4-5): github_context, list_label_wrapper, list_workflows_wrapper, list_pull_requests — prefer the mcpscripts wrappers over native get_label/list_workflows where available; they eliminate the constant _meta.serverInfo icon overhead (~2.2KB/~560 tokens per call) that otherwise dominates small responses.
• Tools needing improvement: get_teams (grant permission or drop from rotation), list_issues (surface why an item was redacted rather than a terse notice), list_code_scanning_alerts (severity filter should accept a list, per its own guidance, but the live schema only accepts one enum value).
• Context-efficient tools (low tokens, high rating): github_context (75 tokens, zero MCP cost), list_label_wrapper (106 tokens), list_pull_requests (104 tokens).
• Context-heavy tools: get_file_contents (871 tokens) and list_discussions (696 tokens) — both pay a fixed ~2.2KB icon tax per call regardless of payload size; batching multiple reads per session would amortize this better than one-off calls.

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Trend

Size vs Usefulness

Generated by 🔍 GitHub MCP Structural Analysis · agent · 101.9 AIC · ⌖ 13.4 AIC · ⊞ 12.2K · ◷

• expires on Aug 21, 2026, 4:13 AM UTC-08:00