Query, convert, and validate JSON/YAML/TOML files via mcp-json-yaml-toml (uvx)

Recipe: query, convert, and validate structured config files via mcp-json-yaml-toml

Surface

• Package: mcp-json-yaml-toml (PyPI)
• Version: 3.4.2
• Transport: stdio
• Launch: uvx mcp-json-yaml-toml
• Auth: none (fully local, credential-free)
• Tools: 8 — data (get/set/delete), data_query (yq expressions), data_convert, data_merge, data_diff, data_schema, constraint_validate, constraint_list

What it does

Gives agents a safe, structured interface for reading, querying, converting, and validating JSON, YAML, and TOML files. Instead of grep/sed/regex (which break indentation and hallucinate keys), agents get:

• Key-path access: data(file, "get", key_path="database.pool_size") → 10
• yq queries: data_query(file, ".features | keys") → ["dark_mode", "beta_api", "max_upload_mb"]
• Format conversion: data_convert(yaml_file, "json") → pretty-printed JSON
• Schema validation: data_schema("validate", file) → auto-discovers schemas from SchemaStore.org
• Deep merge, diff: Compare or merge two config files across formats

Verified trace (4 calls, all succeeded)

1. Get a nested value by key path:

→ tools/call  data  {file_path: "config.yaml", operation: "get", key_path: "database.pool_size"}
← {success: true, result: "10\n", format: "yaml"}

2. Convert YAML → JSON:

→ tools/call  data_convert  {file_path: "config.yaml", output_format: "json"}
← {success: true, input_format: "yaml", output_format: "json",
   result: '{"server":{"host":"0.0.0.0","port":8080,"workers":4},"database":{"url":"postgres://localhost:5432/mydb","pool_size":10,"timeout_ms":5000},"features":{"dark_mode":true,"beta_api":false,"max_upload_mb":50}}'}

3. Query with yq expression:

→ tools/call  data_query  {file_path: "config.yaml", expression: ".features | keys"}
← {success: true, result: "- dark_mode\n- beta_api\n- max_upload_mb\n"}

4. Validate package.json against auto-discovered schema:

→ tools/call  data_schema  {action: "validate", file_path: "package.json"}
← {syntax_valid: true, schema_validated: true, schema_message: "Schema validation passed", overall_valid: true}

MCP config

{
  "mcpServers": {
    "jyt": {
      "command": "uvx",
      "args": ["mcp-json-yaml-toml"]
    }
  }
}

Failure modes observed

• None in this run. All 4 calls returned success: true with correct data.
• Schema validation auto-discovers schemas from local VS Code extensions and SchemaStore catalogs — if neither is available, schema_validated may be false with an advisory message (not an error).

When to reach for this

• Agent needs to read a specific config value without loading the entire file into context
• Converting between config formats (e.g., migrating YAML→TOML for a Rust project)
• Validating that a config file conforms to its schema before committing
• Comparing two config files (e.g., staging vs production) with structured diff
• Setting a single value in a deeply nested config without risk of breaking the file