Model Merger -- MCP Tool Specification

Version: v1.1.9 | Status: Active | Last Updated: March 2026

Overview

This document defines the MCP tools exposed by the model_merger module. These tools are auto-discovered by the PAI MCP bridge via the @mcp_tool decorator in mcp_tools.py and surfaced as part of the ~303 dynamic tools available to Claude.

Auto-Discovery

Property	Value
Discovery method	@mcp_tool decorator scan
Namespace	model_merger
Trust default	Safe
PAI bridge	src/codomyrmex/agents/pai/mcp/

Tool Reference

merge_models

Description: Merge two model parameter sets using SLERP or linear interpolation. Trust Level: Safe Category: analysis

Parameters:

Parameter	Type	Required	Default	Description
params_a	dict[str, list]	Yes	--	First model parameters (key -> list of floats)
params_b	dict[str, list]	Yes	--	Second model parameters (key -> list of floats)
method	str	No	"slerp"	Merge method ('slerp' or 'linear')
alpha	float	No	0.5	Interpolation weight in [0, 1]. 0=model A, 1=model B

Returns: dict[str, Any] -- Dictionary with merged parameter keys and shapes, or error info.

Example:

from codomyrmex.model_merger.mcp_tools import merge_models

result = merge_models(
    params_a={"layer1": [1.0, 2.0, 3.0]},
    params_b={"layer1": [4.0, 5.0, 6.0]},
    method="linear",
    alpha=0.5,
)

---

create_model_soup

Description: Create a model soup by averaging multiple model parameter sets. Trust Level: Safe Category: analysis

Parameters:

Parameter	Type	Required	Default	Description
param_dicts	list[dict[str, list]]	Yes	--	List of model parameter dictionaries (key -> list of floats)
weights	list[float]	No	None	Optional weighting for each model (uniform if omitted)

Returns: dict[str, Any] -- Dictionary with result parameter keys, shapes, and model count.

Example:

from codomyrmex.model_merger.mcp_tools import create_model_soup

result = create_model_soup(
    param_dicts=[
        {"layer1": [1.0, 2.0]},
        {"layer1": [3.0, 4.0]},
        {"layer1": [5.0, 6.0]},
    ],
    weights=[0.5, 0.3, 0.2],
)

Integration Notes

• Auto-discovered: Yes (via @mcp_tool in mcp_tools.py)
• Trust Gateway: All tools are safe -- no destructive operations
• PAI Phases: BUILD (model combination), VERIFY (merge validation)
• Dependencies: Requires numpy and internal merger module

Navigation

• README.md | AGENTS.md | SPEC.md