[mcp-analysis] MCP Structural Analysis - 2026-08-14

[github-actions[bot]]

Daily structural analysis of GitHub MCP tool responses across 10 representative tools (one per toolset). This is day 3 of the 30-day rolling trend window.

Executive Summary

Metric	Value
Tools Analyzed	10
Total Tokens (Today)	13,932
Average Usefulness Rating	3.6/5
Best Rated Tool	github_context (workflow-injected): 5/5
Worst Rated Tool	get_teams: 1/5

Recommendations

• High-value tools (rating 4–5): github_context (injected), list_pull_requests, list_workflows (mcpscripts wrapper), list_code_scanning_alerts, list_discussions, get_label, search_code.
• Tools needing improvement: get_teams (permission never granted across all 3 sampled days — needs an explicit grant, or should be swapped out as the context toolset representative), get_file_contents (no way to fetch a section/range of a large file, so README reads keep growing as an unbounded contributor list expands).
• Context-efficient tools (low tokens, high rating): list_pull_requests (30 tokens, 4/5), list_code_scanning_alerts (10 tokens, 4/5), github_context (90 tokens, 5/5).
• Context-heavy tools: get_file_contents (11,800 tokens today) — driven almost entirely by an unbounded, ever-growing community-contributor link list embedded in the README body, not by the file's actual documentation content.

Full Structural Analysis Report

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Assessment
github_context (injected)	workflow_context	⭐⭐⭐⭐⭐	Excellent — zero MCP overhead, immediately actionable identity data
list_pull_requests	pull_requests	⭐⭐⭐⭐	Clean, minimal, field-filtered
list_workflows (mcpscripts wrapper)	actions	⭐⭐⭐⭐	perPage honored correctly, no icon bloat
list_code_scanning_alerts	code_security	⭐⭐⭐⭐	Clean scoped empty-array result
list_discussions	discussions	⭐⭐⭐⭐	Good structure, cursor pagination present
get_label	labels	⭐⭐⭐⭐	Flat, efficient single-item lookup
search_code	search	⭐⭐⭐⭐	Well-scoped, actionable
get_file_contents	repos	⭐⭐⭐	Usable but bloated by unrelated contributor list
list_issues	issues	⭐⭐⭐	Usable, but integrity-policy redaction hides completeness
get_teams	context	⭐	Blocked by ungranted permission — zero data for 3rd straight day

Schema Analysis

Tool	Type	Depth	Key Fields	Notes
github_context	markdown	1	actor, repository, workspace, workflow-run-id	Prompt-injected, no MCP call
get_file_contents	object	4	content, resource.text, resource.uri, _meta.serverInfo	Deep nesting from resource wrapper
list_issues	array	2	number, title, state	Flat, field-filtered
list_pull_requests	array	2	number, title, state	Flat, field-filtered
list_workflows	object	3	total_count, workflows[].id/name/path/state/html_url	Array of objects under a count wrapper
list_code_scanning_alerts	array	1	(none — empty)	Flat
list_discussions	object	3	discussions[].category.name, title, user.login, pageInfo, totalCount	Cursor pagination present
get_label	object	1	color, description, id, name	Flat
search_code	object	3	total_count, items[].name, items[].path	Flat when text_matches omitted
get_teams	error	0	—	Permission error, no schema

Response Size Analysis

Toolset	Avg Tokens (3-day)	Tools Tested
repos	9,200	3
labels	1,483	3
actions	1,217	3
discussions	850	3
search	787	3
workflow_context	90	3
issues	72	3
pull_requests	28	3
code_security	10	3
context	10	3

Tool-by-Tool Analysis (Today, 2026-08-14)

Tool	Toolset	Tokens	Schema	Rating	Notes
github_context	workflow_context	90	markdown	5	Stable identity block
get_teams	context	15	error	1	Permission never granted, 3rd day blocked
get_file_contents	repos	11,800	object	3	README keeps growing via contributor list
list_issues	issues	65	array	3	1 issue redacted by integrity policy
list_pull_requests	pull_requests	30	array	4	Clean single-PR result
list_workflows	actions	172	object	4	Wrapper fix holds day-over-day
list_code_scanning_alerts	code_security	10	array	4	Empty, clean
list_discussions	discussions	650	object	4	totalCount 9,540, up from 9,496
get_label	labels	550	object	4	Single-label lookup efficient
search_code	search	550	object	4	1,022 total matches, scoped result

30-Day Trend Summary

Metric	Value
Data Points	30 (3 days × 10 tools)
Average Daily Tokens	13,747
Average Rating Trend	Improving (3.3 → 3.6 → 3.6)

Methodology

Each day, one representative tool per GitHub MCP toolset is called with minimal parameters (perPage=1, single-item lookups, field filters where supported) against github/gh-aw. Response size is estimated at ~4 characters/token. Usefulness is rated 1–5 on completeness, actionability, clarity, efficiency, and relationship data. Results are appended to a rolling 30-day JSONL log in cache memory and charted with pandas/matplotlib/seaborn.

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Trend

Size vs Usefulness

References:

• §31798871381

Generated by 🔍 GitHub MCP Structural Analysis · agent · 121.2 AIC · ⌖ 47.9 AIC · ⊞ 11.8K · ◷

• expires on Aug 15, 2026, 4:21 AM UTC-08:00

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-08-15T12:21:22.342Z.

Closed by Workflow