Tencent/WeKnora · error

rerank model is not configured: please set rerank_model_id o

Error message

rerank model is not configured: please set rerank_model_id on the agent

What it means

When the agent has the knowledge_search tool enabled, AgentQA requires a rerank model. It reads req.CustomAgent.Config.RerankModelID; if the agent needs reranking (agentRequiresRerankModel) and the field is empty, this error is returned before retrieval. Wiki-only agents that don't need reranking are exempt.

Source

Thrown at internal/application/service/session_agent_qa.go:111

	// Get rerank model from custom agent config only when knowledge_search can
	// actually run. A disabled KB scope makes all KB tools ineffective, so it
	// must not force users to configure an otherwise-unused rerank model.
	var rerankModel rerank.Reranker
	if agentRequiresRerankModel(req.CustomAgent) {
		// Rerank model is resolved purely from the agent config now.
		// We used to fall back to ConversationConfig.RerankModelID at
		// the tenant level, but that path encouraged "leave rerank
		// blank on the agent and inherit silently" which made debugging
		// retrieval quality a guessing game across tenant settings vs
		// agent settings. Forcing the agent to declare its own rerank
		// model puts the configuration where the user actually edits
		// the agent. If a Wiki-only agent doesn't need reranking,
		// agentRequiresRerankModel() below already lets it pass.
		rerankModelID := req.CustomAgent.Config.RerankModelID
		if rerankModelID == "" {
			logger.Warnf(ctx, "No rerank model configured for custom agent %s, but knowledge_search tool is enabled", req.CustomAgent.ID)
			return errors.New("rerank model is not configured: please set rerank_model_id on the agent")
		}

		rerankModel, err = s.modelService.GetRerankModel(ctx, rerankModelID)
		if err != nil {
			logger.Warnf(ctx, "Failed to get rerank model: %v", err)
			return fmt.Errorf("failed to get rerank model: %w", err)
		}
	} else {
		logger.Infof(ctx, "knowledge_search is unavailable for the effective agent scope, skipping rerank model initialization")
	}

	// Load multi-turn history directly from DB (the single source of truth).
	// AgentSteps on each historical assistant message are expanded into proper
	// assistant_with_tool_calls + tool messages so the model can see what was
	// tried last turn — except final_answer, which is replayed as the trailing
	// canonical assistant message.
	var llmContext []chat.Message
	if agentConfig.MultiTurnEnabled {

View on GitHub (pinned to 988cbb0330)

Solutions

1. Set rerank_model_id on the agent's config to a valid rerank model ID.
2. Alternatively disable the knowledge_search tool if reranking is not needed.
3. Validate via modelService.GetRerankModel that the configured rerank model exists.

Example fix

// before
cfg := &types.AgentConfig{Tools: []string{"knowledge_search"}} // no RerankModelID
// after
cfg := &types.AgentConfig{Tools: []string{"knowledge_search"}, RerankModelID: "bge-reranker-v2-m3"}

if slices.Contains(agent.Config.Tools, "knowledge_search") && agent.Config.RerankModelID == "" {
    return fmt.Errorf("agent %s enables knowledge_search but has no rerank_model_id", agent.ID)
}

func needsRerankModel(a *types.CustomAgent) bool {
    return slices.Contains(a.Config.Tools, "knowledge_search") && a.Config.RerankModelID == ""
}

When it happens

Trigger: Calling AgentQA with a CustomAgent that enables the knowledge_search tool but leaves Config.RerankModelID empty.

Common situations: Agents enabled for knowledge search after creation without adding a rerank model, cloned agents whose rerank_model_id was not copied, or rerank model removed from the registry while the field was cleared.

Understand the failure class

Background: "X is required", "must be set", "cannot be empty": the missing-required-config error family, from Vertex AI project/location to WeChat keys — this error's family across 18 libraries.

Related errors

• custom agent configuration is required for agent QA
• summary model (model_id) is not configured in custom agent s
• invalid sandbox type
• timeout cannot be negative
• memory limit cannot be negative

AI-assisted analysis of Tencent/WeKnora@988cbb0330 (2026-09-02). Data as JSON: /api/errors/a092b9ae9e0380ca. Report an issue: GitHub.