Skip to content

feat(agent-config): add 10 new LLM provider families to dropdown and config UI - #162

Open
prasadadhav wants to merge 2 commits into
BESSER-PEARL:developfrom
prasadadhav:feat/llm_support_for_baf
Open

feat(agent-config): add 10 new LLM provider families to dropdown and config UI#162
prasadadhav wants to merge 2 commits into
BESSER-PEARL:developfrom
prasadadhav:feat/llm_support_for_baf

Conversation

@prasadadhav

@prasadadhav prasadadhav commented Jul 16, 2026

Copy link
Copy Markdown
Contributor

Summary

  • Extends AgentLLMProviderType (editor package) and AgentLLMProvider (webapp
    shared types) union types with all 10 new provider keys.
  • Adds labelled dropdown entries to AGENT_LLM_PROVIDER_OPTIONS in
    AgentConfigurationPanel (e.g. "Google (Gemini)", "xAI (Grok)", "Meta (Llama)").
  • Updates normalizeAgentLLMElement and the local-storage validation whitelist to
    accept all new keys (array-based check, replaces hardcoded conditionals).
  • Extends isChatCompatibleProvider in agent-state-update.tsx to return true for
    all new families (all support chat-completion).
  • Adds per-provider form fields, YAML serializer stanzas, and collapsible <Section>
    components in AgentConfigYamlEditor for each new provider's api_key.
  • Retains legacy 'huggingfaceapi' (no underscore) alongside canonical
    'huggingface_api' to avoid breaking existing saved configs.

Related PRs:

@ArmenSl

ArmenSl commented Jul 20, 2026

Copy link
Copy Markdown
Collaborator

Frontend wiring looks clean and the provider keys match the backend (#575) exactly. Same two gaps as the paired backend PR — flagging here for tracking, both were the bar on the Ollama PR:

1. No docs. The provider reference in docs/source/generators/agent_personalization.rst (BESSER repo) enumerates the supported providers (llmProvider: openai, huggingface, huggingfaceapi, replicate) and their per-provider model lists, so it's now actively incomplete with the 10 new families. That update belongs in the paired #575 — noting it here so the feature doesn't merge without it.

2. No tests. No coverage for the new providers on this side either. The Ollama PR added a round-trip test on the backend; over here a small test exercising normalizeAgentLLMElement / the local-storage provider whitelist / the dropdown options for a couple of the new keys would lock in that they round-trip and don't silently fall back to openai.

Otherwise LGTM.

@prasadadhav

Copy link
Copy Markdown
Contributor Author

Hi Armen :)
I added the docs and tests for the new LLM famailies.
Hope this is what you meant. Or let me know what else needs to be done.

@prasadadhav
prasadadhav force-pushed the feat/llm_support_for_baf branch from b70a218 to b875a25 Compare July 21, 2026 07:16
…families

Add Mistral, DeepSeek, Google (Gemini), Meta (Llama), Anthropic (Claude), Qwen,
xAI (Grok), Groq, Together AI, and OpenRouter to all provider-key locations:
AgentLLMProviderType union, AgentLLMProvider type, AgentConfigurationPanel dropdown
options and normalization whitelist, isChatCompatibleProvider predicate, and
AgentConfigYamlEditor form fields with YAML serializer stanzas and collapsible UI
sections per provider.
…d VALID_AGENT_LLM_PROVIDERS module-level const (used internally by normalizeAgentRuntimeConfig)

- AgentConfigurationPanel.tsx: added export to AGENT_LLM_PROVIDER_OPTIONS, normalizeAgentLLMElement, AgentLLMElementProvider, and AgentLLMElement
- New agent-llm.test.ts: 48 parametrized cases covering all three things the reviewer asked for normalizeAgentLLMElement preserves new keys without falling back to openai, dropdown options completeness, and local-storage whitelist completeness. 48/48 pass. (The 1 pre-existing failure in modifiers.test.ts is unrelated to our changes.)
@prasadadhav
prasadadhav force-pushed the feat/llm_support_for_baf branch from b875a25 to 41daf80 Compare July 21, 2026 07:27
@ArmenSl

ArmenSl commented Jul 22, 2026

Copy link
Copy Markdown
Collaborator

Nice perfect, when it's released in BAF I will release it here

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants