[mcp-analysis] MCP Structural Analysis - 2026-08-03

[github-actions[bot]]

MCP Structural Analysis — 2026-08-03

Tested 11 representative tools across GitHub MCP toolsets. Average usefulness 3.36/5. github_context, list_discussions, and list_pull_requests (field-scoped) rated best (⭐⭐⭐⭐⭐); get_teams remains unusable (permission denied) and list_workflows/list_issues/search_repositories showed known structural problems (unhonored pagination, integrity-filtered results, and a mis-scoped repo: qualifier).

Metric	Value
Tools Analyzed	11
Total Tokens (Today)	13,938
Average Usefulness Rating	3.36/5
Best Rated	github_context, list_discussions, list_pull_requests: 5/5
Worst Rated	get_teams: 1/5

Full Structural Analysis Report

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Assessment
github_context	workflow_context	⭐⭐⭐⭐⭐	Zero-cost prompt-injected identity block, immediately actionable
list_discussions	discussions	⭐⭐⭐⭐⭐	Compact, cursor pagination works, embeds category+author
list_pull_requests (fields-scoped)	pull_requests	⭐⭐⭐⭐⭐	Field projection cuts payload to near-zero, fully actionable
get_label	labels	⭐⭐⭐⭐	Flat 4-field object, zero bloat once name is known
get_file_contents	repos	⭐⭐⭐⭐	Clean resource envelope, but no partial/range fetch — cost scales with file size
list_code_scanning_alerts	code_security	⭐⭐⭐⭐	Minimal payload on empty result, ideal for polling
search_users	users	⭐⭐⭐	Adequate but minimal fields (no name/bio/company)
list_workflows	actions	⭐⭐	per_page param silently ignored — always returns ~30 of 345 workflows
list_issues	issues	⭐⭐	Integrity policy silently filters results to [], misleading agent into believing zero issues exist
search_repositories	search	⭐⭐	repo: qualifier not honored — returned an unrelated high-star repo instead of the target
get_teams	context	⭐	Permission not granted under the read-only integration token — unusable in CI

Schema Analysis (Today)

Tool	Type	Depth	Key Fields	Notes
github_context	markdown	1	actor, repository, workspace, run-id	Prompt-injected, no MCP call
get_teams	error	0	—	Permission denial
get_file_contents	object	3	content[].text(SHA), resource.uri/mimeType/text	MCP resource envelope
list_issues	array	1	(filtered)	Integrity-redacted, empty array shown
list_pull_requests	array	1	number, title, state, html_url	Field-scoped
list_workflows	object	2	total_count, workflows[].id/name/path/state/html_url/badge_url	Unbounded page
list_code_scanning_alerts	array	0	—	Empty (no critical alerts)
list_discussions	object	3	discussions[].category/title/user/html_url, pageInfo, totalCount	Cursor pagination
get_label	object	1	name, color, description, id	Flat lookup
search_users	object	2	total_count, items[].login/id/profile_url/avatar_url	Minimal identity
search_repositories	array	1	full_name, stargazers_count, forks_count, topics	Query scoping unreliable

Response Size Analysis — 30-Day Average by Toolset

Toolset	Avg Tokens
repos	5,328
actions	2,610
code_security	1,669
pull_requests	1,407
issues	1,008
search	603
labels	223
discussions	182
users	136
workflow_context	76
context	34

Tool-by-Tool Analysis (Today, 2026-08-03)

Tool	Toolset	Tokens	Schema	Rating	Notes
github_context	workflow_context	75	markdown	5	run=30815299944, all trigger numbers #false (scheduled)
get_teams	context	22	error	1	"you haven't granted it yet" — consistent across ~15 runs
get_file_contents (README.md)	repos	9,600	object	4	300+ community-contributor links dominate size
list_issues	issues	65	array	2	1 item redacted by integrity policy (#49985)
list_pull_requests	pull_requests	48	array	5	Clean single result (#49972)
list_workflows	actions	3,700	object	2	per_page=1 ignored, returned 30 of 345
list_code_scanning_alerts	code_security	5	array	4	Empty — no open critical alerts
list_discussions	discussions	155	object	5	totalCount=9,121 (up from 9,020 on 07-31, ~+50/day)
get_label	labels	38	object	4	'bug' label lookup
search_users	users	90	object	3	query='pelikhan'
search_repositories	search	140	array	2	repo: qualifier mis-scoped — returned unrelated github/gitignore two runs in a row (also seen 07-31)

30-Day Trend Summary

Metric	Value
Data Points	174
Date Range	2026-07-06 to 2026-08-03
Average Daily Tokens	~14,277
Rating Trend	Stable — same structural issues (unhonored pagination on list_workflows, integrity-filtered list_issues/search_issues, get_teams permission denial) reproduce consistently across the window

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Trend

Size vs Usefulness

Recommendations

• High-value tools (rating 4–5): github_context, list_discussions, field-scoped list_pull_requests, get_label, get_file_contents, list_code_scanning_alerts
• Tools needing improvement: list_workflows (respect per_page/perPage), list_issues/search_issues (surface a visible flag when integrity policy redacts results instead of silently returning []), search_repositories (honor repo: qualifier instead of ranking by popularity)
• Context-efficient + high value: github_context (75 tok, r5), list_discussions (155 tok, r5), get_label (38 tok, r4)
• Context-heavy: get_file_contents on large files (~9.6k tok today), list_workflows (~3.7k tok, capped by an ignored pagination param)

References:

• §30815299944

Generated by 🔍 GitHub MCP Structural Analysis · agent · 104.6 AIC · ⌖ 48.8 AIC · ⊞ 11.6K · ◷

• expires on Aug 4, 2026, 5:05 AM UTC-08:00

[github-actions[bot]]

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

A newer discussion is available at Discussion #50256.