Consolidate duplicate Venice AI client implementations into single canonical service

[devJaja]

Summary

There are currently two separate Venice AI client implementations in the backend:

1. backend/src/services/venice/ — the canonical, well-tested client with cache, dedup, and circuit breaker (client.ts, cache.ts, dedup.ts)
2. backend/src/venice/ — a smaller duplicate with its own circuitBreaker.ts and client.ts

This causes: divergent retry/backoff behavior, duplicated API key handling, confusion about which client agents use, and maintenance burden. The smaller duplicate also appears to be missing cache/dedup functionality.

Priority

P0 — Two competing Venice clients cause divergent behavior, duplicated config, and untracked fixes.

Files to work on

backend/src/services/venice/ (canonical — keep)
backend/src/venice/ (delete or merge)
backend/src/services/venice/circuitBreaker.ts (restore if missing)

Requirements

• Audit usage of both clients across backend/src/agents/ and backend/src/coordinator/
• Consolidate all imports to backend/src/services/venice/
• Migrate any unique functionality from backend/src/venice/ into backend/src/services/venice/
• Delete backend/src/venice/ directory
• Update all imports and verify no residual references remain
• Add a test that verifies only one VeniceClient class exists

Acceptance Criteria

• Only backend/src/services/venice/ exists — backend/src/venice/ is deleted
• All agents and coordinator import from backend/src/services/venice/
• No duplicate imports of VeniceClient from both paths
• grep -r "from '../venice/" returns no matches in backend/src/
• All existing Venice client tests pass
• npm run lint and npm test pass

[AngelAnselmoRS]

I have already investigated the codebase and fully resolved this refactoring. I consolidated the duplicate Venice AI client modules into backend/src/services/venice/, updated all import paths, cleaned up the obsolete directory, added a regression guard test, and opened a Pull Request with the solution.

[AngelAnselmoRS]

Clarification: This merged PR addresses and resolves issue #491 (Consolidate duplicate Venice AI client implementations into single canonical service).

[AngelAnselmoRS]

Hi @devJaja! GrantFox support mentioned that because the PR was merged before manual assignment, it can't be updated through Discord support. Could you please check on your GrantFox maintainer dashboard if you can trigger the assignment/reward distribution for issue #491? Thank you!