fix(state): rebuild legacy gateway_routing PK; guard session_store in dispatch hook

[teknium1]

Summary

Heals live databases stuck with the pre-scope gateway_routing PRIMARY KEY, ending per-save "routing save failed" warning spam, and guards the pre_gateway_dispatch hook's session_store access so the hook always fires.

Root cause: early builds of the #59203 routing-index migration created gateway_routing with session_key TEXT PRIMARY KEY and no scope column. The declarative column reconciler ADDs the missing scope column, but SQLite cannot ALTER a primary key — so the shipped composite PRIMARY KEY (scope, session_key) never lands on those databases and every routing save fails:

• save_gateway_routing_entry → ON CONFLICT clause does not match any PRIMARY KEY or UNIQUE constraint
• replace_gateway_routing_entries → UNIQUE constraint failed: gateway_routing.session_key whenever the same session_key exists under another scope

One live errors.log carried 261 of these warnings; the store silently falls back to sessions.json every time.

Changes

• hermes_state.py: new _heal_gateway_routing_pk() — detects the legacy PK via PRAGMA table_info, rebuilds the table once with the composite key, preserves rows (newest wins on collisions, NULL scope coalesced to ''). Wired into _init_schema() right after _reconcile_columns(). Same one-time-heal pattern as the [Bug]: Gateway Memory Loss — INSERT omits active column in hermes_state.py #51646 active-column repair.
• gateway/run.py: pre_gateway_dispatch kwargs now pass getattr(self, "session_store", None) — a partially-initialized runner previously raised inside the dispatch try-block, logging 'GatewayRunner' object has no attribute 'session_store' and dropping the hook for every message.
• Tests: TestGatewayRoutingPkHeal (4 cases: rebuild, both write paths post-heal, NULL-scope coalesce, idempotence on modern shape) + hook-fires-without-session_store regression. All fail without their fixes (sabotage-verified).

Validation

	Before	After
Legacy-PK DB, upsert	ON CONFLICT error, warning per save	succeeds
Legacy-PK DB, cross-scope replace	UNIQUE constraint error	succeeds, scopes isolated
Real affected state.db (copy)	261 warnings/log	heals on open, 15 live rows preserved
Runner without session_store	hook never fires, warning per message	hook fires with store=None

scripts/run_tests.sh tests/test_hermes_state.py tests/gateway/test_session.py tests/gateway/test_session_store_lock_io.py → 626 passed, 0 failed. tests/gateway/test_pre_gateway_dispatch.py → 6 passed.

Infographic

[github-actions]

૮ >ﻌ< ა ci review

_{ran on 49e13fd}

ℹ️ Info

Desktop E2E visual evidence · View test artifacts · View job

1 visual diff.

_{inline evidence upload failed.}

Failed to upload diff-665a0833239e-onboarding-overlay-diff.png with gh image (exit code 1): Error uploading /home/runner/work/_temp/e2e-evidence/diff-665a0833239e-onboarding-overlay-diff.png: step 0 (get upload token): uploadToken not found on repo page — do you have write access to NousResearch/hermes-agent? (or, if NousResearch enforces SAML SSO, authorize at https://github.com/orgs/NousResearch/sso)