Tool Index Format

ToolStorePy tool indexes should be treated as format-agnostic. For documentation and ecosystem use, JSON is the preferred source format because it is easier to validate, transform, publish, and integrate with other systems.

The underlying retrieval requirement is simple: each tool record needs structured metadata that can later be chunked and embedded.

Preferred JSON shape

[
  {
    "tool_id": "1",
    "tool_name": "calculator",
    "tool_description": "Securely evaluate arithmetic expressions.",
    "tool_git_link": "https://github.com/example/calculator.git"
  }
]

Core fields

Field	Meaning
tool_id	Unique identifier for the tool entry
tool_name	Human-readable label used in output and search results
tool_description	Retrieval text that should describe the capability well
tool_git_link	Repository URL that will later be cloned by the build pipeline

Example entry

{
  "tool_id": "1",
  "tool_name": "calculator",
  "tool_description": "This is a secure mathematical calculator designed to evaluate a text-based arithmetic expression.",
  "tool_git_link": "https://github.com/example/calculator.git"
}

Why JSON is preferred

• it is already a standard interchange format
• it is easier to validate with schemas
• it avoids custom escaping rules
• it is simpler for contributors and downstream tooling

Metadata chunking

ToolStorePy does not embed raw repository source code when authoring an index. It embeds a chunked metadata representation like this:

ID: 1
Name: calculator
Description: This is a secure mathematical calculator designed to evaluate a text-based arithmetic expression.
Git Link: https://github.com/example/calculator.git

That distinction matters. Search quality depends heavily on the quality of tool_description.

Legacy note

Some source-project examples still use a .toon file and script names such as embed_toon.py. Treat that as an implementation detail of the current reference pipeline rather than the preferred long-term index format.