Tools Overview¶
Tools are callable functions the LLM can invoke to take real-world actions. Each tool has a clear name, typed parameters, and a docstring description so Claude knows exactly when and how to use it.
All Tools at a Glance¶
| Tool | Category | Description |
|---|---|---|
factorial_safe_forLLM |
Basic | Returns n! with safe error handling |
echo |
Basic | Echoes back any message โ great for connectivity checks |
add_note |
Storage | Appends text to a local notes file |
read_notes |
Storage | Reads all saved notes |
get_cryptocurrency_price |
API | Fetches live crypto prices via CoinGecko |
perform_web_search |
API | Searches the web via Perplexity/OpenAI |
capture_screenshot |
Visual | Takes a screenshot and returns it as JPEG |
add_person_to_member_database |
Storage | Logs a structured Person record |
save_memory |
Memory | Saves text to an OpenAI vector store |
search_memory |
Memory | Semantic search over saved memories |
How Tools Work¶
sequenceDiagram
participant U as User
participant C as Claude
participant S as MCP Server
participant X as External Service
U->>C: "What's the Bitcoin price?"
C->>S: Call get_cryptocurrency_price("bitcoin")
S->>X: GET api.coingecko.com
X-->>S: { "bitcoin": { "usd": 67000 } }
S-->>C: "The current price of bitcoin is $67,000 USD."
C-->>U: Friendly answer with context
Copy-paste test prompts
See the ๐ฌ Chat Prompt Recipes section for ready-made prompts to test every tool.