CLI Reference

toolstorepy build

toolstorepy build --queries <path> [--index <name> | --index-url <url>] [options]

Flags

Flag	Default	Description
--queries	required	Path to queries.json with tool_description entries
--index	—	Name of a built-in index such as core-tools
--index-url	—	URL to a downloadable index archive
--workspace	toolstorepy_workspace	Output workspace directory
--install-requirements	off	Install cloned repo requirements into the workspace venv
--host	0.0.0.0	Host the generated MCP server binds on
--port	8000	Port the generated MCP server listens on
--llm-scan	off	Enable LLM-based autonomous security review (bypasses human prompt)
--llm-model	claude-sonnet-4-6	Model for LLM scanning — any LangChain-supported string
--force-refresh	off	Re-download the index archive even if cached
--verbose	off	Enable verbose logging and full tracebacks on failure

Examples

# Basic build
toolstorepy build --queries queries.json --index core-tools

# Custom port
toolstorepy build --queries queries.json --index core-tools --port 9090

# Custom host and port
toolstorepy build --queries queries.json --index core-tools --host 127.0.0.1 --port 8080

# LLM-based autonomous security scan with default model (Claude)
export ANTHROPIC_API_KEY=sk-...
toolstorepy build --queries queries.json --index core-tools --llm-scan

# LLM scan with GPT-4o
export OPENAI_API_KEY=sk-...
toolstorepy build --queries queries.json --index core-tools --llm-scan --llm-model gpt-4o

# LLM scan with Gemini
export GOOGLE_API_KEY=...
toolstorepy build --queries queries.json --index core-tools --llm-scan --llm-model gemini-2.0-flash

# Full build with all options
toolstorepy build \
  --queries queries.json \
  --index core-tools \
  --workspace ./my_workspace \
  --install-requirements \
  --host 0.0.0.0 \
  --port 9090 \
  --llm-scan \
  --verbose

---

toolstorepy cache

Populate

Pre-warm the bare-repo cache from a resolved queries file (items must have a git_link field) or from explicit URLs.

# From a resolved queries file
toolstorepy cache populate --queries resolved.json

# From explicit URLs (can be repeated)
toolstorepy cache populate --url https://github.com/org/repo1.git \
                           --url https://github.com/org/repo2.git

# Force re-cache even if already cached
toolstorepy cache populate --url https://github.com/org/repo.git --force

List

toolstorepy cache list

Clear

toolstorepy cache clear

Cache command caveat

cache populate --queries expects items that already include git_link fields, not tool_description. That differs from the normal build input format. Use --url for direct URLs, or run a full toolstorepy build first to resolve descriptions to URLs and warm the cache automatically.

---

Server configuration

--host and --port are baked into the generated mcp_unified_server.py at build time. The output file will contain:

if __name__ == "__main__":
    mcp.run(transport='streamable-http', host='0.0.0.0', port=8000)

The transport is always streamable-http. Only host and port are configurable.

---

LLM scan model strings

--llm-model accepts any LangChain-supported model string. The provider is inferred automatically from well-known prefixes.

Provider	Example model string	Required env var	Install
Anthropic	claude-sonnet-4-6	ANTHROPIC_API_KEY	pip install langchain-anthropic
OpenAI	gpt-4o	OPENAI_API_KEY	pip install langchain-openai
Google	gemini-2.0-flash	GOOGLE_API_KEY	pip install langchain-google-genai
Mistral	mistral-large-latest	MISTRAL_API_KEY	pip install langchain-mistralai

You can also use explicit provider prefix syntax: anthropic:claude-sonnet-4-6, openai:gpt-4o.

---

Index authoring scripts

These are part of the current reference index-authoring workflow.

Create tool metadata interactively

python vector_db_creation/toon_file_creation.py

Build the vector database

python vector_db_creation/embed_toon.py

---

Typical build commands

# Minimal
toolstorepy build --queries queries.json --index core-tools

# With requirements installed and custom port
toolstorepy build --queries queries.json --index core-tools --install-requirements --port 9090

# Fully automated with LLM security scan (no human prompts)
toolstorepy build --queries queries.json --index core-tools --llm-scan