Jai/conv

[J-SUPHA]

What does this PR do?

Adds a per-session conversation=<session_id> tag to Nous Portal inference
requests so Portal-side usage can be attributed to a specific Hermes session.

Every Portal request already carries two product-attribution tags
(product=hermes-agent, client=hermes-client-v<version>) built centrally in
agent/portal_tags.py. The Nous provider profile already receives the agent's
session_id but discarded it. This PR threads that id into the tag list as an
additive, opt-in third tag, emitted only when a session id is present.

The approach keeps the single-source-of-truth pattern in agent/portal_tags.py
(no inlined literals) and is fully backward compatible: nous_portal_tags()
with no argument still returns the canonical two-tag list, so auxiliary/base
call sites are unchanged.

Related Issue

N/A (no tracking issue). Happy to open one if maintainers prefer.

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/portal_tags.py
  • Added conversation_tag(session_id: str) -> str returning
    conversation=<session_id> (mirrors the existing hermes_client_tag()
    factoring; documents that this tag is high-cardinality and opt-in).
  • Extended nous_portal_tags(session_id: str | None = None) to append the
    conversation tag only when a truthy session_id is provided. Default
    behavior (no arg) is unchanged → still
    ["product=hermes-agent", "client=hermes-client-v<version>"].
• plugins/model-providers/nous/__init__.py
  • NousProfile.build_extra_body() now forwards its session_id kwarg into
    the tag list (alongside the existing provider_preferences handling).
• tests/agent/test_portal_tags.py
  • Added: conversation-tag format, presence when a session id is supplied,
    and omission when it is None/"".
• tests/providers/test_provider_profiles.py
  • Added TestNousProfile.test_tags_include_conversation_when_session_id.
• tests/run_agent/test_provider_parity.py
  • Updated the end-to-end _build_api_kwargs Nous parity assertion to expect
    the conversation tag when the agent has a session_id.

Scope / known limitations (intentional)

• The tag rides the main agent loop only. Auxiliary/background Portal calls
  (compression, title generation, vision, web-extract, session search, kanban,
  curator, etc.) build tags without a session id and remain untagged, so
  per-conversation cost analytics built on this tag will undercount.
• session_id rotates on context compression and /new, so a single
  user-facing conversation can emit multiple conversation= values. The value
  is the current session segment, not a stable end-to-end conversation id.
• Only the opaque session_id (<timestamp>_<uuid>) is sent — never the
  gateway session key, which can embed PII (phone numbers / user ids).

How to Test

1. Point Hermes at the Nous Portal (hermes model → provider nous, or set
   model.provider: nous in ~/.hermes/config.yaml).

2. Confirm the profile emits the tag when a session id is present:

python -c "
from providers import get_provider_profile
p = get_provider_profile('nous')
print('no session :', p.build_extra_body())
print('w/ session :', p.build_extra_body(session_id='conv-abc-123'))
"

Expected:

no session : {'tags': ['product=hermes-agent', 'client=hermes-client-v<ver>']}
w/ session : {'tags': ['product=hermes-agent', 'client=hermes-client-v<ver>', 'conversation=conv-abc-123']}

3. Run the targeted tests:

pytest tests/agent/test_portal_tags.py \
       tests/providers/test_provider_profiles.py \
       tests/run_agent/test_provider_parity.py::TestBuildApiKwargsNousPortal -q

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 (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: macOS (Apple Silicon), Python 3.13

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — docstrings in agent/portal_tags.py updated; no user-facing docs needed
• I've updated cli-config.yaml.example if I added/changed config keys — N/A (no new config keys)
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — N/A
• I've considered cross-platform impact (Windows, macOS) — N/A (pure Python, no OS-specific code)
• I've updated tool descriptions/schemas if I changed tool behavior — N/A

Screenshots / Logs

$ pytest tests/agent/test_portal_tags.py tests/providers/test_provider_profiles.py \
         tests/run_agent/test_provider_parity.py::TestBuildApiKwargsNousPortal -q
.....................................................................    [100%]
69 passed in 1.14s

[teknium1]

Merged via PR #65468 with your commits cherry-picked onto current main — your authorship is preserved in git history. Thanks for the clean, well-tested contribution!

We extended your work in the same PR to cover the known limitations you called out in the description: a ContextVar-based ambient conversation context now propagates the conversation= tag to auxiliary calls (compression, titles, vision, MoA advisor slots), and the tag value is the session-lineage ROOT id, so one conversation keeps a single stable value across context-compression rotation. Delegate subagent trees attribute to their parent conversation as well.