[mcp-analysis] MCP Structural Analysis - 2025-12-08

[github-actions[bot]]

Today's analysis covers 10 GitHub MCP tools across 7 toolsets. The average usefulness rating for agentic work is 4.4/5, with most tools rated 4-5 stars. Context-efficient tools like list_workflows, list_discussions, and get_label provide high value with minimal token usage (35-305 tokens). The get_me tool remains inaccessible due to permission constraints.

Full Structural Analysis Report

Executive Summary

Metric	Value
Tools Analyzed	10
Total Tokens (Today)	9,075
Average Usefulness Rating	4.4/5
Best Rated Tools	get_file_contents, list_issues, list_workflows, list_discussions, get_label, list_commits, list_branches: 5/5
Worst Rated Tool	get_me: 1/5

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Assessment
get_file_contents	repos	⭐⭐⭐⭐⭐	Excellent for autonomous agents - returns clean file content as resource
list_issues	issues	⭐⭐⭐⭐⭐	Excellent - comprehensive issue data with all needed context
list_workflows	actions	⭐⭐⭐⭐⭐	Excellent - perfect balance of completeness and efficiency
list_discussions	discussions	⭐⭐⭐⭐⭐	Excellent - clean structure with pagination support
get_label	labels	⭐⭐⭐⭐⭐	Excellent - minimal, focused data with no bloat
list_commits	repos	⭐⭐⭐⭐⭐	Excellent - well-structured commit data, clean and actionable
list_branches	repos	⭐⭐⭐⭐⭐	Excellent - minimal, efficient branch data
list_pull_requests	pull_requests	⭐⭐⭐⭐	Good - detailed PR data but verbose with some redundancy
search_code	search	⭐⭐⭐⭐	Good - rich search results but context-heavy
get_me	context	⭐	Poor - inaccessible due to permissions

Schema Analysis

Tool	Type	Depth	Key Fields	Notes
get_file_contents	resource	1	content, sha, encoding	Clean resource format
list_issues	object_with_array	4	issues, pageInfo, totalCount, number, title, state, labels, user	Nested user/label objects
list_pull_requests	array	5	number, title, state, user, head, base, assignees, draft	Deep nesting with repo objects
list_workflows	object_with_array	2	total_count, workflows, id, name, path, state	Compact structure
list_discussions	object_with_array	3	discussions, pageInfo, totalCount, number, title, category	Category embedded
get_label	object	1	color, description, id, name	Flat structure
search_code	object_with_array	5	total_count, items, name, path, repository	Embedded repo objects
list_commits	array	3	sha, commit, author, committer, message	Logical nesting
list_branches	array	1	name, sha, protected	Minimal structure
get_me	error	0	-	403 error response

Response Size Analysis

Toolset	Avg Tokens	Tools Tested
pull_requests	3,650	1 (list_pull_requests)
search	1,750	1 (search_code)
repos	705	3 (get_file_contents, list_commits, list_branches)
issues	920	1 (list_issues)
actions	305	1 (list_workflows)
discussions	275	1 (list_discussions)
labels	35	1 (get_label)
context	25	1 (get_me - error)

Tool-by-Tool Analysis

Tool	Toolset	Tokens	Schema	Rating	Notes
list_pull_requests	pull_requests	3,650	array (depth 5)	4	Rich but verbose, some redundant fields
search_code	search	1,750	object_with_array (depth 5)	4	Full repo context included, context-heavy
get_file_contents	repos	1,575	resource (depth 1)	5	Clean resource format, immediately actionable
list_issues	issues	920	object_with_array (depth 4)	5	Complete issue context, excellent for agents
list_commits	repos	480	array (depth 3)	5	Well-structured, logical nesting
list_workflows	actions	305	object_with_array (depth 2)	5	Compact and complete, ideal for agents
list_discussions	discussions	275	object_with_array (depth 3)	5	Clean with pagination, excellent structure
list_branches	repos	60	array (depth 1)	5	Minimal and efficient, perfect for agents
get_label	labels	35	object (depth 1)	5	No bloat, just essential data
get_me	context	25	error (depth 0)	1	403 error, not usable in this context

30-Day Trend Summary

Metric	Value
Data Points	90
Average Daily Tokens	8,482
Average Rating Trend	stable
Date Range	2025-11-26 to 2025-12-08

Recommendations

Based on the analysis:

High-value tools (rating 4-5):

• Context-efficient champions: get_label (35 tokens), list_branches (60 tokens), list_discussions (275 tokens), list_workflows (305 tokens)
• Rich but efficient: list_commits (480 tokens), list_issues (920 tokens)
• Resource access: get_file_contents (1,575 tokens) - essential for file operations
• Good with trade-offs: search_code (1,750 tokens), list_pull_requests (3,650 tokens)

Tools needing improvement:

• get_me - Permission issues make it unusable with GitHub Actions tokens

Context-efficient tools (low tokens, high rating):

• list_workflows, list_discussions, get_label, list_commits, list_branches - All provide excellent value under 500 tokens

Context-heavy tools (high tokens):

• list_pull_requests (3,650 tokens) - Consider using with pagination or filtering
• search_code (1,750 tokens) - Rich results justified by search functionality

Best practices for agentic workflows:

1. Use get_label, list_branches, list_workflows for lightweight context
2. Leverage list_issues for comprehensive issue analysis
3. Use get_file_contents for targeted file access
4. Apply pagination with list_pull_requests to control token usage
5. Avoid get_me in GitHub Actions contexts (use context toolset alternatives)

Visualizations

Response Size by Toolset

Average token usage varies significantly by toolset. Pull requests are the most token-intensive, while labels and context tools are the most efficient.

Usefulness Ratings

Most toolsets achieve excellent ratings (4-5 stars) for agentic work. Only the context toolset scores low due to permission constraints.

Daily Token Trend

Token usage remains stable over the 30-day period, indicating consistent response sizes across toolsets.

Size vs Usefulness

Most tools cluster in the high-usefulness range (4-5 stars) regardless of size. The most efficient tools (small size, high rating) are in the bottom-right: get_label, list_branches, list_workflows, list_discussions, and list_commits.

AI generated by GitHub MCP Structural Analysis

[github-actions[bot]]

⚓ Avast! This discussion be marked as outdated by GitHub MCP Structural Analysis.
🗺️ A newer treasure map awaits ye at Discussion #5938.
Fair winds, matey! 🏴‍☠️