[mcp-analysis] MCP Structural Analysis - 2026-09-07

[github-actions[bot]]

Analyzed 10 representative GitHub MCP tool calls across 9 toolsets today. Best-rated: github_context and list_workflows (mcpscripts wrapper) at ⭐⭐⭐⭐⭐; worst: list_issues at ⭐⭐ due to a redacted item hiding all visible content. Notable finding: switching the repos test target from .gitignore to README.md roughly 4.5x'd that tool's response size, driven entirely by an auto-generated 250+-entry community-contributor list embedded in the file — a reminder that get_file_contents cost is target-dependent, not fixed overhead.

Full Structural Analysis Report

Executive Summary

Metric	Value
Tools Analyzed	10
Total Tokens (Today)	14,832
Average Usefulness Rating	3.60/5
Best Rated Tool	github_context / list_workflows: 5/5
Worst Rated Tool	list_issues: 2/5

Usefulness Ratings for Agentic Work

Tool	Toolset	Rating	Assessment
github_context	workflow_context	⭐⭐⭐⭐⭐	Zero-cost injected identity block, always actionable
list_workflows (mcpscripts)	actions	⭐⭐⭐⭐⭐	Clean JSON, explicit pagination, no icon bloat
list_pull_requests	pull_requests	⭐⭐⭐⭐	Flat, minimal, no overhead
search_code	search	⭐⭐⭐⭐	Inline code fragment avoids a follow-up call
list_code_scanning_alerts	code_security	⭐⭐⭐⭐	Rich remediation detail, but inconsistent schema across scanners
get_file_contents	repos	⭐⭐⭐	Correct content, but icon/_meta overhead + target-dependent size
list_discussions	discussions	⭐⭐⭐	Good pagination metadata, icon overhead dominates small payloads
get_label	labels	⭐⭐⭐	Tiny real payload buried in fixed icon overhead
search_users	users	⭐⭐⭐	Minimal identity fields only, needs follow-up for profile detail
list_issues	issues	⭐⭐	Integrity-filtered item hides all content, blocks assessment

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	3	content.resource.text, _meta.serverInfo	Fixed ~2.2k-char base64 icon block on every call
list_issues	array	2	number, title, state, labels	Redacted item shows zero content
list_pull_requests	array	1	number, title, state, html_url	No icon/_meta overhead
list_workflows (wrapper)	object	2	total_count, workflows[].id/name/path/state	Clean wrapper, explicit pagination
list_code_scanning_alerts	array	3	rule.id, rule.full_description, rule.help, most_recent_instance.location	Schema shape identical but field population varies wildly by scanning tool
list_discussions	object	3	discussions[].category.name, pageInfo.hasNextPage, totalCount	Nested category/user relationships
get_label	object	1	name, color, description, id	Flat but dwarfed by icon overhead
search_users	object	2	items[].login, items[].id, total_count	No name/bio/role fields
search_code	object	2	items[].path, items[].text_matches[].fragment	Fragment inlines actual matching code

Response Size Analysis

Toolset	Avg Tokens	Tools Tested
repos	2,456	9
code_security	1,651	9
labels	778	9
discussions	683	9
search	616	9
users	610	9
actions	170	9
workflow_context	98	9
issues	61	9
pull_requests	38	9

(Averages across all 9 measured days, 2026-08-25 to 2026-09-07)

Tool-by-Tool Analysis (Today, 2026-09-07)

Tool	Toolset	Tokens	Schema	Rating	Notes
github_context	workflow_context	95	markdown	⭐⭐⭐⭐⭐	Zero MCP cost
get_file_contents	repos	7,500	object	⭐⭐⭐	README.md target ballooned size vs typical .gitignore reads
list_issues	issues	75	array	⭐⭐	1 item redacted, 9th consecutive day of this pattern
list_pull_requests	pull_requests	30	array	⭐⭐⭐⭐	Flat, tiny
list_workflows	actions	172	object	⭐⭐⭐⭐⭐	mcpscripts wrapper, no bloat
list_code_scanning_alerts	code_security	4,300	array	⭐⭐⭐⭐	10 alerts across 4 distinct scanner tools
list_discussions	discussions	675	object	⭐⭐⭐	Icon overhead dominates
get_label	labels	610	object	⭐⭐⭐	Icon overhead dominates
search_users	users	650	object	⭐⭐⭐	Identity-only fields
search_code	search	725	object	⭐⭐⭐⭐	Inline fragment

30-Day Trend Summary

Metric	Value
Data Points	90
Average Daily Tokens	7,161
Average Rating Trend	Stable (~3.3–3.7/5 across all 9 measured days, no drift)

Recommendations

• High-value tools (rating 4-5): github_context, list_workflows (mcpscripts wrapper), list_pull_requests, search_code, list_code_scanning_alerts. Prefer these when the task allows.
• Tools needing improvement: list_issues — the integrity-policy redaction has now hidden content on all 9 measured days; consider whether the redacted item's metadata (at least type/state) could surface without full content. list_code_scanning_alerts — schema is inconsistent across third-party scanning tools (CodeQL populates full remediation fields; Semgrep/Malicious-Code-Scanner/Smoke-Claude leave rule.name/description/help empty and cram detail into message instead), forcing agents to branch on tool name rather than a uniform field.
• Context-efficient tools (low tokens, high rating): list_pull_requests (30 tokens), list_workflows via mcpscripts wrapper (172 tokens) — both avoid the fixed ~2.2k-char base64 icon _meta.serverInfo block that inflates repos/discussions/labels/users responses.
• Context-heavy tools: get_file_contents (repos) — cost is highly target-dependent; today's README.md pull was ~4.5x a typical .gitignore read due to an auto-generated contributor list. list_code_scanning_alerts — scales directly with open alert count (ranged 10 tokens to 5,200+ tokens across measured days) and severity filter chosen.

Visualizations

Response Size by Toolset

Usefulness Ratings

Daily Token Trend

Size vs Usefulness

References: §34119036176

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• api.anthropic.com

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "api.anthropic.com"

See Network Configuration for more information.

Generated by 🔍 GitHub MCP Structural Analysis · claude · agent · 186.5 AIC · ⌖ 13.6 AIC · ⊞ 11.5K · ◷

• expires on Sep 8, 2026, 4:03 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 #59451.