Loading ToolBench

Reading the current corpus…

mlb-mcp MCP server · F · ToolBench

← Browse

mlb-mcp

github.com/etweisberg/mlb-mcp

An MCP server providing MLB statistics and baseball analysis tools using MLB StatsAPI and pybaseball data

2026-07-28+STREAMABLE_HTTP0 tools

26/100

Definition Quality0

Protocol Readiness0

Supportability43

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence · evidence: stateless requests, Streamable HTTP

Current-spec patterns detected

Summary

Scoring was not performed

Issues (7)

HIGHtool-description ↗

Descriptions lack WHEN/WHY context. 'Create a strike zone plot...' does not explain when an agent should choose this tool vs alternatives, what prerequisites exist, or what the user gets back. Descriptions are 40-80 chars, below the 50-200 char target for LLM-optimized docs.

create_strike_zone_plotcreate_spraychart_plotcreate_bb_profile_plotcreate_teams_plot

HIGHconstrained-input ↗

Parameters lack constraints. 'colorby' and 'annotation' accept free-form strings with no enum declaration. An LLM may pass 'pitch_colored_by_type' or other hallucinated values. No mention of valid column names from the Statcast data schema.

create_strike_zone_plotcreate_spraychart_plotcreate_bb_profile_plotcreate_teams_plot

HIGHrecovery-guide ↗

Error handling is basic exception wrapping ('Error creating strike zone plot: {e}') without recovery guidance. When plotting fails due to bad column names or missing data, the LLM has no suggestion for what to do next (e.g., 'Try with parameter=exit_velocity' or 'Call get_statcast_data first').

Recommendations

Expand tool descriptions to 100-180 characters. Answer: What does it do? When do I call it? What will I get? Example: 'Create a scatter plot of pitch locations overlaid on the strike zone. Call this after fetching Statcast pitch data (use get_statcast). Returns a PNG image and pitch statistics.'
Add enum constraints to 'colorby', 'annotation', 'parameter', and 'x_axis'/'y_axis' parameters. Document valid Statcast column names (pitch_type, exit_velocity, launch_angle, events, etc.) or provide a discovery tool (list_statcast_columns) agents can call first.
Replace generic error messages with recovery guidance. Example: 'Column "foo" not found in Statcast data. Available columns: pitch_type, exit_velocity, launch_angle, hc_x, hc_y, events. Try parameter=exit_velocity or call list_statcast_columns().'
Add min/max bounds to numeric parameters. Example: 'size (int, 10-500): marker size in pixels, default 100' and 'width (int, 200-2000): plot width in pixels, default 500'.
Document the structure of the 'data' input parameter in the schema. Specify that it must be a dict with a 'data' key containing a list of pitch/hit/team records, or simplify the interface to accept the list directly.
Include chaining references in output. E.g., if create_strike_zone_plot returns a plot, also return the data_source or athlete_id so downstream tools know what was plotted.
Add a dependency hint to descriptions: 'Prerequisite: Call get_statcast_data() or search_player() first to populate the data parameter.'

Score history

Overall score trend

↓ 20 points across a rubric change (v1 → v2)

26/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-22	F	26	2026-07-28+	v2
2026-03-09	F	46	-	v1