[mcp-analysis] MCP Structural Analysis - 2026-07-03

[github-actions[bot]]

GitHub MCP Structural Analysis — 2026-07-03

Analyzed 10 representative tools across 10 toolsets on github/gh-aw (server awmg-github v0.3.32, 54 tools exposed). Total today: 7,662 tokens, average usefulness 3.9/5. The 30-day rating trend is improving (3.39 → 3.83).

Key findings

• 🏆 Best-in-class (5/5): get_label, list_discussions, search_repositories. Flat/compact schemas, and list_discussions is still the only tool with real GraphQL cursor pagination (pageInfo + totalCount).
• ✅ search_repositories jumped from 3,969 → 117 tokens once perPage=1 was set — confirming the standing recommendation: always bound page size.
• ⚠️ Context-heavy: get_file_contents (4,635 tok, full README inline) and list_issues (1,799 tok even at perPage=1) — cost is body-driven, not schema bloat.
• 🚫 get_me 403s under the workflow integration token — normally 5/5, unusable in this CI identity; read actor from workflow context instead.
• 🧭 actions_list is a dispatcher requiring a method arg (e.g. list_workflows) or it errors — an extra step agents must know.

Recommendations

• High-value (4–5): list_discussions, get_label, search_repositories, search_users, list_pull_requests, list_issues, actions_list.
• Context-efficient (low tokens, high rating): get_label (40), search_users (65), search_repositories (117), list_discussions (158).
• Use with care (bound/scope them): get_file_contents (target small files), list_issues (expect long AI-generated bodies), search_repositories (never omit perPage).
• Blocked this identity: get_me — do not rely on it in workflows.

Full Structural Analysis Report

Executive Summary

Metric	Value
Tools Analyzed	10
Total Tokens (Today)	7,662
Average Usefulness Rating	3.9/5
Best Rated Tools	get_label / list_discussions / search_repositories: 5/5
Worst Rated Tool	get_me: 1/5 (403 under integration token)

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Assessment
get_label	labels	⭐⭐⭐⭐⭐	Flat, complete, minimal — ideal
list_discussions	discussions	⭐⭐⭐⭐⭐	Only true cursor pagination + totalCount
search_repositories	search	⭐⭐⭐⭐⭐	All discovery signals; efficient when perPage set
list_pull_requests	pull_requests	⭐⭐⭐⭐	Very complete; body + embedded repos inflate cost
list_issues	issues	⭐⭐⭐⭐	Actionable core fields; long bodies dominate cost
actions_list	actions	⭐⭐⭐⭐	Clean once method convention is known
search_users	users	⭐⭐⭐⭐	Compact handle-resolution; no deep cursor
get_file_contents	repos	⭐⭐⭐⭐	Full file content; cost scales with file size
list_code_scanning_alerts	code_security	⭐⭐⭐	Correct-but-empty this run; schema not observable
get_me	context	⭐	403 under integration token — unusable in CI

Schema Analysis

Tool	Type	Depth	Key Fields	Notes
get_label	object	1	name, color, description, id	Flat, cleanest schema
get_me	object	2	error, isError	Error envelope (403)
get_file_contents	object	2	content(text), path	Decoded file inline
actions_list	object	2	total_count, workflows[]	Dispatcher; needs method
list_discussions	object	3	discussions[], pageInfo, totalCount	Cursor pagination
search_users	object	3	total_count, items[].login/id/urls	Envelope, no cursor
search_repositories	array	2	full_name, stars, forks, topics, language	Discovery signals
list_issues	array	3	number, title, state, labels, user, body	labels[]/user{} nesting
list_pull_requests	array	4	number, state, draft, merged, head, base	head/base embed repo
list_code_scanning_alerts	array	1	(empty)	[] — no open critical alerts

Response Size Analysis

Toolset	Avg Tokens	Tool Tested
repos	4,635	get_file_contents
issues	1,799	list_issues
pull_requests	677	list_pull_requests
discussions	158	list_discussions
actions	132	actions_list
search	117	search_repositories
users	65	search_users
labels	40	get_label
context	38	get_me (error)
code_security	1	list_code_scanning_alerts

Tool-by-Tool Analysis

Tool	Toolset	Tokens	Schema	Rating	Notes
get_file_contents	repos	4,635	object/2	4	Full README inline; cost ∝ file size
list_issues	issues	1,799	array/3	4	Long AI-generated bodies dominate
list_pull_requests	pull_requests	677	array/4	4	Complete; embedded repo redundancy
list_discussions	discussions	158	object/3	5	pageInfo + totalCount=8005
actions_list	actions	132	object/2	4	Requires method arg
search_repositories	search	117	array/2	5	perPage=1 → 117 vs 3969
search_users	users	65	object/3	4	total_count, no cursor
get_label	labels	40	object/1	5	Flat, complete
get_me	context	38	object/2	1	403 integration token
list_code_scanning_alerts	code_security	1	array/1	3	Empty (no open critical)

30-Day Trend Summary

Metric	Value
Data Points	162
Sampling Days	18
Average Daily Tokens	~7,407
Overall Avg Rating	3.48/5
Rating Trend	Improving (3.39 → 3.83)

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Trend

Size vs Usefulness

References: §28660619938

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• awmgmcpg

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "awmgmcpg"

See Network Configuration for more information.

Generated by 🔍 GitHub MCP Structural Analysis · 177.2 AIC · ⌖ 16.3 AIC · ⊞ 9.9K · ◷

• expires on Jul 4, 2026, 4:44 AM UTC-08:00