[mcp-analysis] MCP Structural Analysis - January 23, 2026

[github-actions[bot]]

Comprehensive quantitative and qualitative analysis of GitHub MCP tool responses evaluating size, schema structure, and usefulness for agentic workflows.

Executive Summary

Metric	Value
Tools Analyzed	13
Usable Tools	11
Error/Unavailable	2
Total Tokens (Today)	32,878
Average Tokens per Tool	2,989
Average Usefulness Rating	3.77/5

Top Performers:

• Best Rated: get_file_contents (repos) - 5/5, 1,315 tokens
• Most Efficient: get_label (labels) - 5/5, 28 tokens
• Needs Improvement: list_code_scanning_alerts (code_security) - 2/5, 24,215 tokens

Usefulness Rating Distribution

Rating	Assessment	Count
⭐⭐⭐⭐⭐ 5/5	Excellent for autonomous agents	6 tools
⭐⭐⭐⭐ 4/5	Good, minor improvements possible	3 tools
⭐⭐⭐ 3/5	Adequate, requires supplementary calls	1 tool
⭐⭐ 2/5	Limited usefulness	1 tool
⭐ 1/5	Unusable for agentic tasks	2 tools

Full Structural Analysis Report

Toolset Rankings by Usefulness

Toolset	Avg Rating	Avg Tokens	Nesting Depth	Tools Tested
actions	⭐⭐⭐⭐⭐ 5.00	180	2.0	1
discussions	⭐⭐⭐⭐⭐ 5.00	240	3.0	1
labels	⭐⭐⭐⭐⭐ 5.00	28	1.0	1
repos	⭐⭐⭐⭐ 4.75	779	1.8	4
issues	⭐⭐⭐⭐ 4.00	1,580	4.0	1
search	⭐⭐⭐⭐ 4.00	320	3.0	1
pull_requests	⭐⭐⭐ 3.00	3,200	5.0	1
code_security	⭐⭐ 2.00	24,215	5.0	1
context	⭐ 1.00	ERROR	0.0	1
notifications	⭐ 1.00	ERROR	0.0	1

Schema Analysis by Tool

Tool	Toolset	Tokens	Schema Type	Nesting	Rating	Key Fields
get_label	labels	28	object	1	⭐⭐⭐⭐⭐	id, name, color, description
list_branches	repos	52	array	1	⭐⭐⭐⭐⭐	name, sha, protected
list_tags	repos	68	array	2	⭐⭐⭐⭐⭐	name, commit, zipball_url, tarball_url
list_workflows	actions	180	object	2	⭐⭐⭐⭐⭐	total_count, workflows
list_discussions	discussions	240	object	3	⭐⭐⭐⭐⭐	discussions, pageInfo, totalCount
search_repositories	search	320	object	3	⭐⭐⭐⭐	total_count, incomplete_results, items
get_file_contents	repos	1,315	string	1	⭐⭐⭐⭐⭐	content, encoding, sha, size
list_issues	issues	1,580	object	4	⭐⭐⭐⭐	issues, pageInfo, totalCount
get_repository_tree	repos	1,680	object	3	⭐⭐⭐⭐	sha, tree, truncated
list_pull_requests	pull_requests	3,200	array	5	⭐⭐⭐	number, title, user, head, base, assignees
list_code_scanning_alerts	code_security	24,215	array	5	⭐⭐	number, rule, tool, most_recent_instance, state
get_me	context	ERROR	error	0	⭐	N/A
list_notifications	notifications	ERROR	error	0	⭐	N/A

Detailed Assessments

⭐⭐⭐⭐⭐ Excellent Tools (Rating 5/5)

get_file_contents (repos) - 1,315 tokens

• Schema: string, Depth 1
• Assessment: Perfect file retrieval - returns base64 content with metadata, clean structure, immediately actionable for file operations
• Recommendation: Ideal for agents needing file access

list_workflows (actions) - 180 tokens

• Schema: object, Depth 2
• Assessment: Excellent minimal workflow listing - flat structure with only essential fields (id, name, path, state), perfect efficiency
• Recommendation: Context-efficient workflow discovery

list_discussions (discussions) - 240 tokens

• Schema: object, Depth 3
• Assessment: Clean discussion listing with GraphQL pagination - includes category, user, timestamps, well-balanced metadata without bloat
• Recommendation: Well-designed for discussion management

get_label (labels) - 28 tokens

• Schema: object, Depth 1
• Assessment: Perfect minimal label data - flat structure with only core fields, excellent context efficiency
• Recommendation: Most efficient tool tested today

list_branches (repos) - 52 tokens

• Schema: array, Depth 1
• Assessment: Perfect minimal branch listing - flat structure with only essential data, excellent efficiency
• Recommendation: Excellent for branch operations

list_tags (repos) - 68 tokens

• Schema: array, Depth 2
• Assessment: Clean tag listing with commit reference - well-balanced metadata for release management
• Recommendation: Good balance of data and efficiency

⭐⭐⭐⭐ Good Tools (Rating 4/5)

list_issues (issues) - 1,580 tokens

• Schema: object, Depth 4
• Assessment: Good issue listing with GraphQL pagination - includes nested user/label objects with full details, slightly verbose but comprehensive
• Recommendation: Comprehensive but could be lighter

search_repositories (search) - 320 tokens

• Schema: object, Depth 3
• Assessment: Good search results with minimal_output=true - concise repo objects without excessive metadata, well-optimized for discovery
• Recommendation: Well-optimized search implementation

get_repository_tree (repos) - 1,680 tokens

• Schema: object, Depth 3
• Assessment: Good tree structure with file/dir entries - includes type, mode, sha for each entry, useful for file exploration but can be large for big repos
• Recommendation: Consider pagination for large trees

⭐⭐⭐ Adequate Tools (Rating 3/5)

list_pull_requests (pull_requests) - 3,200 tokens

• Schema: array, Depth 5
• Assessment: Very heavy response - full repository objects nested in head/base, complete user objects for assignees/reviewers, useful but bloated with redundant repo metadata
• Recommendation: Consider minimal mode or field filtering

⭐⭐ Limited Usefulness (Rating 2/5)

list_code_scanning_alerts (code_security) - 24,215 tokens

• Schema: array, Depth 5
• Assessment: Extremely verbose - 96KB for alert listing, includes full rule definitions and location details for each alert, needs summary mode for efficiency
• Recommendation: Critical - needs summary mode or pagination improvements

⭐ Unusable Tools (Rating 1/5)

get_me (context) - ERROR

• Issue: 403 permission error - not accessible by integration token
• Recommendation: Use alternative user context methods

list_notifications (notifications) - ERROR

• Issue: 403 permission error - not accessible by integration token
• Recommendation: Not available for workflow automation

30-Day Trend Summary

Metric	Value
Data Points Collected	167
Analysis Days	14
Average Daily Tokens	82,240
Date Range	2026-01-05 to 2026-01-23

Recommendations

High-Value Tools (Rating 4-5, Context-Efficient)

Use these tools frequently for efficient agentic workflows:

1. get_label (28 tokens) - Minimal label data
2. list_branches (52 tokens) - Efficient branch listing
3. list_tags (68 tokens) - Clean tag metadata
4. list_workflows (180 tokens) - Minimal workflow listing
5. list_discussions (240 tokens) - Well-balanced discussion data
6. search_repositories (320 tokens) - Optimized search results

Context-Heavy Tools Requiring Optimization

list_code_scanning_alerts - 24,215 tokens

• Issue: Includes full rule definitions and location details for each alert
• Impact: Consumes 74% of today's total token budget
• Recommendation: Implement summary mode or field filtering

Permission Issues

get_me and list_notifications both return 403 errors with integration tokens. These tools are unavailable for GitHub Actions workflows and require alternative approaches:

• Use repository context instead of user context
• Check workflow trigger events for notification-like information

Schema Complexity

Tools with nesting depth ≥4 may be harder for agents to parse:

• list_pull_requests (Depth 5) - Full repo objects in head/base
• list_code_scanning_alerts (Depth 5) - Full rule definitions
• list_issues (Depth 4) - Nested user/label objects

Consider flattening or providing minimal modes for these tools.

Methodology

This analysis evaluated GitHub MCP tools across three dimensions:

1. Quantitative: Response size in tokens (1 token ≈ 4 characters)
2. Structural: Schema type, nesting depth, key fields
3. Qualitative: Usefulness rating (1-5) for agentic workflows

Rating Criteria:

• 5: Complete, actionable data with clear structure
• 4: Most needed data present, minor gaps
• 3: Usable but requires additional calls
• 2: Missing key data, hard to parse
• 1: Minimal value for agentic tasks

---

Workflow Run: §21283952178

AI generated by GitHub MCP Structural Analysis

• expires on Jan 30, 2026, 11:10 AM UTC

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-01-30T11:10:49.781Z.