Explicit Tool Calls

Write code. Call APIs. No guessing.

OneTool's __ot prefix gives you explicit control over tool invocation. You write the code - the LLM doesn't have to guess which tool or which parameters.

Trigger Prefixes

OneTool supports five trigger prefixes. All prefixes support the same three invocation styles.

Prefix	Description	Recommended
__onetool__run	Full name, explicit tool call	Yes
__ot	Short form	Yes
__onetool	Full name, default tool
__ot__run	Short name, explicit tool call
mcp__onetool__run	Explicit MCP call

Recommended: Use __onetool__run for clarity or __ot for brevity.

Note: mcp__ot__run is NOT a valid prefix.

Invocation Styles

Each prefix supports three ways to pass code:

1. Simple Call

Direct function call after the prefix:

__ot sha256(text="hello")
__onetool__run multiply(a=8472, b=9384)

2. Inline Backticks

Code wrapped in backticks:

__ot `sha256(text="hello")`
__onetool__run `multiply(a=8472, b=9384)`

3. Code Fence

Multi-line code in a fenced block:

__onetool__run
```python
metals = ["Gold", "Silver"]
results = {}
for metal in metals:
    results[metal] = brave.web_search(query=f"{metal} price")
results

## Direct MCP Call

For programmatic or explicit MCP invocation:

mcp__onetool__run(command='sha256(text="hello")')

## Complete Examples

### Simple Hash Computation

__ot sha256(text="hello world")

### Multi-step Computation with Variables

__onetool__run

msg = "Hello World"
sha256(text=msg)

### Loop with Multiple Tool Calls

__ot

primes = [is_prime(n=i) for i in range(11, 21)]
primes

### Chained Operations

__ot__run upper(text=reverse(text="hello"))

---

## Prompt Engineering for Tool Calls

### Pre-Call Instructions

Add context before the tool call to guide the LLM:

Calculate the SHA-256 hash of the following text: __onetool__run sha256(text="hello world")

### Post-Call Processing

Request specific formatting or analysis after the tool result:

__onetool__run brave.web_search(query="latest AI news", count=5)

Summarise the top 3 results in bullet points.

### Structured Output Requests

Combine tool execution with output formatting:

    __onetool__run
    ```python
    results = {
        "hash": sha256(text="hello"),
        "reversed": reverse(text="hello"),
        "length": count_chars(text="hello")
    }
    results
    ```

    Return the results as a markdown table.

---

## Best Practices

Your system prompt should include instructions for reliable tool execution:

```yaml
system_prompt: |
  Never retry successful tool calls to get "better" results.
  If a tool call fails, report the error - do not compute the result yourself.

Why these matter:

• No retries on success: LLMs sometimes want to "improve" results by calling the same tool again. This wastes tokens and can cause loops.
• No manual computation on failure: When a tool fails, LLMs often try to compute the answer themselves (e.g., calculating a hash or Fibonacci number). This defeats the purpose of using tools and may produce incorrect results.

---

Troubleshooting

"Syntax error at line 1: invalid syntax"

The prefix wasn't stripped properly. Ensure:

1. There's whitespace or a newline after the prefix
2. The code itself is valid Python

Tool call returns unexpected result

Check that:

1. Function arguments match the expected signature
2. String arguments are properly quoted
3. For multi-line code, the last expression is what you want returned

Code fence not recognized

Ensure:

1. Opening ``` is on its own line after the prefix
2. Closing ``` is on its own line
3. Language hint (e.g., python) is optional but recommended