[mcp-analysis] MCP Structural Analysis - 2026-02-04

[github-actions[bot]]

Analyzed 15 GitHub MCP tools across multiple toolsets for response size, schema structure, and usefulness for agentic workflows. Key findings: list_releases consumes excessive context (18K tokens), while tools like list_branches, list_tags, and search_repositories provide excellent efficiency.

Full Structural Analysis Report

Executive Summary

Metric	Value
Tools Analyzed	15
Total Tokens (Today)	37,380
Average Usefulness Rating	4.1/5
Best Rated Tools	list_branches, list_tags, get_commit, search_repositories, list_discussions: 5/5
Worst Rated Tool	list_releases: 2/5

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Tokens	Assessment
list_branches	repos	⭐⭐⭐⭐⭐	100	Extremely concise - branch name, SHA, protected status
list_tags	repos	⭐⭐⭐⭐⭐	180	Minimal and perfect - tag name and commit SHA
get_commit	repos	⭐⭐⭐⭐⭐	400	Clean metadata without diff bloat
search_repositories	search	⭐⭐⭐⭐⭐	350	Minimal output mode works excellently
list_discussions	discussions	⭐⭐⭐⭐⭐	450	GraphQL-style, very concise and well-structured
get_file_contents	repos	⭐⭐⭐⭐⭐	300	Returns file content with SHA - perfect for agents
list_commits	repos	⭐⭐⭐⭐	600	Well-structured with good balance of detail
list_issues	issues	⭐⭐⭐⭐	4,800	Rich data but can be verbose with full body text
list_workflows	actions	⭐⭐⭐⭐	2,000	Clean workflow list with essential fields
list_label	labels	⭐⭐⭐⭐	4,200	Comprehensive but verbose for many labels
list_pull_requests	pull_requests	⭐⭐⭐	5,600	Contains useful data but bloated with redundant repo info
list_releases	releases	⭐⭐	18,000	Extremely verbose with full asset details - inefficient
get_me	user	⭐	0	403 error - integration token not permitted
list_projects	projects	⭐⭐	0	Error - requires different permissions

Schema Analysis

Tool	Type	Depth	Key Fields	Notes
list_branches	array	1	name, sha, protected	Flat structure
list_tags	array	2	name, commit	Minimal nesting
get_file_contents	string	1	content, sha	Simple message
search_repositories	object	2	total_count, items	Clean pagination
list_discussions	object	3	discussions, pageInfo, totalCount	GraphQL-style
get_commit	object	3	sha, commit, author, committer	Balanced nesting
list_commits	array	3	sha, commit, author, committer	Structured list
list_issues	object	3	issues, pageInfo, totalCount	GraphQL with full bodies
list_workflows	object	2	total_count, workflows	Minimal nesting
list_label	object	2	labels, totalCount	Simple structure
list_pull_requests	array	5	id, number, state, title, user, assignees, head, base	Deep nesting with full repo objects
list_releases	array	5	tag_name, name, body, assets, author	Very deep with asset arrays

Response Size Analysis

Toolset	Avg Tokens	Tools Tested	Context Efficiency
repos	464	5	⭐⭐⭐⭐⭐ Excellent
search	350	1	⭐⭐⭐⭐⭐ Excellent
discussions	450	1	⭐⭐⭐⭐⭐ Excellent
actions	2,000	1	⭐⭐⭐⭐ Good
labels	4,200	1	⭐⭐⭐ Adequate
issues	4,800	1	⭐⭐⭐ Adequate
pull_requests	5,600	1	⭐⭐ Limited
releases	18,000	1	⭐ Poor

Tool-by-Tool Analysis

Tool	Toolset	Tokens	Schema	Rating	Key Observations
list_branches	repos	100	array/1	⭐⭐⭐⭐⭐	Perfect efficiency - just name, SHA, protected flag
list_tags	repos	180	array/2	⭐⭐⭐⭐⭐	Minimal context usage with tag name and commit
get_file_contents	repos	300	string/1	⭐⭐⭐⭐⭐	Returns file with SHA - ideal for agents
search_repositories	search	350	object/2	⭐⭐⭐⭐⭐	Minimal output mode is exceptional
get_commit	repos	400	object/3	⭐⭐⭐⭐⭐	With include_diff=false, returns clean metadata
list_discussions	discussions	450	object/3	⭐⭐⭐⭐⭐	GraphQL-style pagination, very efficient
list_commits	repos	600	array/3	⭐⭐⭐⭐	Good balance of commit info
list_workflows	actions	2,000	object/2	⭐⭐⭐⭐	Clean workflow list with essential fields
list_label	labels	4,200	object/2	⭐⭐⭐⭐	Verbose for repos with many labels (349 in test)
list_issues	issues	4,800	object/3	⭐⭐⭐⭐	Rich data but includes full body text
list_pull_requests	pull_requests	5,600	array/5	⭐⭐⭐	Bloated with full repo objects in head/base
list_releases	releases	18,000	array/5	⭐⭐	Extremely verbose - 15 assets with full uploader info
get_me	user	0	error/0	⭐	403 error with integration token
list_projects	projects	0	error/0	⭐⭐	Requires different auth context

30-Day Trend Summary

Metric	Value
Data Points	30
Days Analyzed	2
Average Daily Tokens	37,380
Trend	Stable (releases increased from 12K to 18K tokens)

Recommendations

High-Value Tools (Rating 5/5, Efficient):

• list_branches - Just 100 tokens for complete branch info
• list_tags - 180 tokens for tag listings
• get_commit - 400 tokens with include_diff=false
• search_repositories - 350 tokens with minimal_output=true
• list_discussions - 450 tokens for discussion listings
• get_file_contents - 300 tokens for file reads

Context-Efficient Pattern:
Use these tools for high-frequency operations where context preservation matters.

Tools Needing Improvement:

• list_releases (18K tokens) - Consider pagination with perPage=1 or filtering
• list_pull_requests (5.6K tokens) - Bloated with redundant repo data in head/base

Context-Heavy Tools (Use Sparingly):

• list_releases - 18K tokens, use only when release details are essential
• list_pull_requests - 5.6K tokens due to full repo objects
• list_label - 4.2K tokens for repos with many labels
• list_issues - 4.8K tokens with full body text

Best Practices for Agents:

1. Prefer repos toolset for file operations (avg 464 tokens)
2. Use minimal_output=true for search operations
3. Set include_diff=false when getting commits
4. Paginate heavily for list_releases (perPage=1)
5. Use GraphQL-based tools (discussions, issues) for efficient pagination

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Trend

Size vs Usefulness

References:

• §21669212414

AI generated by GitHub MCP Structural Analysis

• expires on Feb 11, 2026, 11:18 AM UTC

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-02-11T11:18:21.545Z.

Closed by Workflow