🐛 Bug: LLM.ensure_opg_approval() Rejects Valid min_allowance Values

File: src/opengradient/client/llm.py — LLM.ensure_opg_approval()

What's broken

LLM.ensure_opg_approval() enforces an arbitrary minimum of 0.1 OPG on the min_allowance parameter:

# llm.py line 218-219
if min_allowance < 0.1:
    raise ValueError("min_allowance must be at least 0.1.")

But the underlying opg_token.ensure_opg_approval() function has no such restriction — it accepts any positive value. This creates a split where:

• Small-amount use cases (e.g., min_allowance=0.05) work fine when calling opg_token.ensure_opg_approval() directly
• But calling llm.ensure_opg_approval(min_allowance=0.05) raises a ValueError that's not actually required by the protocol

This 0.1 floor is an undocumented internal constraint with no technical basis, and is inconsistent with PR #201 which changed the hint constant to reference 0.05 OPG.

Fix

Replace the arbitrary floor with a simple positive-number guard (the only invariant that actually matters):

if min_allowance <= 0:
    raise ValueError("min_allowance must be a positive number.")

Updated docstrings to remove references to the obsolete 0.1 minimum.

Impact

Files Changed

• src/opengradient/client/llm.py