PR #17584: Fix curator LLM review to honor curator-specific auxiliary model config

• Repository: NousResearch/hermes-agent
• Author: afurm
• State: open | merged: False
• Link: https://github.com/NousResearch/hermes-agent/pull/17584

Description (problem / solution / changelog)

What does this PR do?

This PR fixes the curator background review fork so it honors curator-specific auxiliary model settings in curator.auxiliary.

Before this change, curator review always resolved its fork runtime from the main model.* config, so it ignored curator.auxiliary.provider and curator.auxiliary.model even when users had explicitly configured a dedicated auxiliary backend.

The fix updates runtime resolution in agent/curator.py to prefer curator overrides when present and fall back to model.* only when curator-specific values are not set.

I also added regression tests in tests/agent/test_curator.py to cover both:

• curator auxiliary override is used,
• and fallback to main model config when curator auxiliary config is missing.

Related Issue

Fixes #17572

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• agent/curator.py: In _run_llm_review(), read curator.auxiliary and prefer curator.auxiliary.provider and curator.auxiliary.model when resolving runtime. Falls back to model.provider / model.default when curator-specific overrides are absent.
• tests/agent/test_curator.py: Added regression tests:
  • test_run_llm_review_uses_curator_auxiliary_config
  • test_run_llm_review_uses_main_config_when_curator_auxiliary_missing

How to Test

1. Configure main model and curator auxiliary settings to different values in ~/.hermes/config.yaml (or through CLI config tooling):
   • model.provider/model.default for primary model
   • curator.auxiliary.provider/curator.auxiliary.model for curator overrides
2. Run:
   • scripts/run_tests.sh tests/agent/test_curator.py -k "run_llm_review_uses_curator_auxiliary"
   • scripts/run_tests.sh tests/agent/test_curator.py -k "run_llm_review_uses_main_config"
3. Confirm tests pass and verify curator metadata in the run summary now reflects curator-specific model/provider when configured.

Checklist

Code

• I’ve read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature
• I’ve run pytest tests/ -q and all tests pass
• I’ve added tests for my changes
• I’ve tested on my platform: <!-- e.g. Ubuntu 24.04, macOS 15.2, Windows 11 -->

Documentation & Housekeeping

• I’ve updated relevant documentation (README, docs/, docstrings) — or N/A
• I’ve updated cli-config.yaml.example if I added/changed config keys — or N/A
• I’ve updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I’ve considered cross-platform impact (Windows, macOS, Linux) — or N/A
• I’ve updated tool descriptions/schemas if I changed tool behavior — or N/A

Changed files

• agent/curator.py (modified, +12/-0)
• tests/agent/test_curator.py (modified, +104/-0)

PR #17591: fix(curator): read curator.auxiliary config for reviewer model

• Repository: NousResearch/hermes-agent
• Author: toller892
• State: closed | merged: False
• Link: https://github.com/NousResearch/hermes-agent/pull/17591

Description (problem / solution / changelog)

Problem

curator.auxiliary.{provider,model} is exposed in the default config (hermes_cli/config.py) but _run_llm_review() only reads cfg["model"], ignoring the curator-specific override. The curator always runs on the main model regardless of what is configured.

Fix

Read curator.auxiliary first; fall back to model config.

Precedence:

1. curator.auxiliary.{provider,model} (explicit curator override)
2. model.{provider,default} (main model config)

Related

Fixes #17572

Changes

• agent/curator.py: +11/-8 lines in _run_llm_review()

Changed files

• agent/curator.py (modified, +11/-8)

PR #17781: fix(curator): honor curator.auxiliary.{provider,model} override (#17572)

• Repository: NousResearch/hermes-agent
• Author: yonefive71
• State: open | merged: False
• Link: https://github.com/NousResearch/hermes-agent/pull/17781

Description (problem / solution / changelog)

Summary

Fixes #17572. The curator's _run_llm_review() resolved provider and model exclusively from cfg["model"], silently ignoring the curator.auxiliary.{provider, model} schema documented in hermes_cli/config.py:950-955. As a result the curator always ran on the user's main model — for Opus users this means weekly runs on the most expensive model when the whole point of an auxiliary slot is to pin a cheaper one.

The bug

agent/curator.py:761-763 (before this patch):

_m = _cfg.get("model", {}) if isinstance(_cfg.get("model"), dict) else {}
_provider = _m.get("provider") or "auto"
_model_name = _m.get("default") or _m.get("model") or ""

No reference to cfg["curator"]["auxiliary"] anywhere in the resolution path, even though the default config block reserves it:

# Optional per-task override for the curator's aux model. Leave null
# to use Hermes' main auxiliary client resolution.
auxiliary:
  provider: null
  model: null

Fix

Read curator.auxiliary first and fall back to the main model fields when either side is null. Provider and model can be overridden independently (provider-only override keeps the main model, and vice versa).

Tests

Three new cases in tests/agent/test_curator.py:

• auxiliary unset → main model (current behavior preserved)
• both fields set → aux wins
• partial override (provider only) → main model retained

pytest tests/agent/test_curator.py → 34 passed.

Notes

• Behavior when both aux fields are null is byte-identical to today.
• resolve_runtime_provider() is still called with the resolved provider+model, so credential pool / OAuth / pool-backed flows continue to work for the aux model.
• No config schema change — the auxiliary block has been in DEFAULT_CONFIG for a while; this just wires it up.

Changed files

• agent/curator.py (modified, +14/-0)
• tests/agent/test_curator.py (modified, +106/-0)