A Go-based MCP server that integrates Claude AI with tools for file operations, Pokémon API queries, and stock ticker data retrieval
This MCP server exposes 6 tools with reasonable naming conventions and descriptions, but significant gaps in parameter documentation, schema completeness, and error handling. All tools have descriptions (meeting a baseline requirement), but most parameters lack formal type constraints, and output schemas are not documented. The tool set mixes file operations, Pokémon API queries, and stock ticker lookups with no clear cohesion. Parameter descriptions are often generic and lack format constraints or constraints on valid values. No evidence of error recovery guidance, permission gating, or audit trails.
Make edits to a text file. Replaces 'old_str' with 'new_str' in the given file. 'old_str' and 'new_str' MUST be different from each other. If the file specified with path doesn't exist, it will be created.
Get ability details by ability name or ID.
Get details of a Pokemon by its name or ID.
Get ticker information for a stock symbol. Use this tool to retrieve real-time or historical stock data based on the provided symbol, period, and interval. Use the stock symbol in the 'symbol' field. The 'period' field can be set to "1d","5d","1mo","3mo","6mo","1y","2y","5y","10y","ytd","max". The 'interval' field can be set to "1m","2m","3m","5m","15m","30m","60m","4h","1d","1wk","1mo","1y". If not specified, defaults are used. If the interval is greater than the period, it will be adjusted to be less than the period.
List files and directories at a given path. If no path is provided, lists files in the current directory.
Ambiguous parameter naming in Pokémon tools: 'key_identifier' and 'value_identifier' are vague and force LLM to guess whether to pass a name, ID, or other value. Parameters should be named by their semantic role (e.g., pokemon_id, pokemon_name) and split into separate parameters or use overloaded resolver pattern.
Missing output schema documentation. No tool documents what fields it returns, forcing LLMs to rely on trial-and-error or implicit assumptions about response structure. Production tools must declare return types (e.g., 'returns array of {id, name, types, abilities, ...}').
No error handling guidance. Tools provide no recovery hints when failures occur (e.g., file not found, Pokémon name invalid, API down). Error responses should guide the LLM toward remediation: 'Pokémon not found. Try calling get_pokemon_detail with a different name or ID.'
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 47 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 47 | - | v1 |
Read the contents of a given relative file path. Use this when you want to see what's inside a file. Do not use this with directory names.
Unconstrained enum parameters in get_ticker: 'period' and 'interval' accept free-form strings with no enum constraint in the JSON Schema. Valid values are documented only in the text description ('1d','5d','1mo',...), which LLMs cannot reliably parse. Use JSON Schema enum to make valid options machine-parseable.
Path traversal vulnerability risk in read_file and edit_file: tools accept relative paths but do not validate against directory traversal (e.g., '../../../etc/passwd'). Tools must sanitize file paths to prevent access outside intended directory.
No permission checking or audit trail. File operations (read_file, edit_file, list_files) perform no access control checks and generate no logs. Production systems require audit trails (who, what, when, result) for compliance and incident response.
Generic parameter descriptions lack format constraints. Example: 'path' in read_file says 'The relative path of a file in the working directory' but does not specify allowed characters, length limits, encoding, or what '..' means. Descriptions should be LLM-actionable: 'A relative file path (ASCII alphanumerics, dots, slashes; no directory traversal; max 255 chars).'
edit_file requires exact string match for replacement but does not document behavior when 'old_str' appears multiple times or does not exist. Descriptions should clarify: 'Fails if old_str does not exist or appears more than once. Returns error with line numbers and context to help the LLM revise.'
Irreversible operations (edit_file, potential delete via editing) lack confirmation or dry-run step. Production tools should support --dry-run or require explicit confirmation for destructive actions to prevent accidental data loss by agents.
list_files does not document pagination. If directories contain thousands of files, the tool may return all of them, exhausting the context window. Even for file operations, clarify: 'Returns at most 100 entries; use limit/offset for large directories.' (Not present in current spec; assume unbounded response).