fix(llm): register LlmChatRunner in LLMAgent worker DI

[ModerRAS]

Symptom

Production log (2026-09-16) shows the agent worker process dying as soon as it picked up a task:

09:59:38 INF  主进程：LLMAgent <chat_id> <pid>已启动
09:59:46 DBG  ChunkPollingService: tracking task <task_id>...
09:59:48 INF  [主进程：LLMAgent <chat_id> <pid>] exited with code -532462766   # 0xE0434352 = unhandled CLR exception
10:00:50 WRN  Recovering LLM agent session for chat <chat_id>: heartbeat timeout

Every LLM chat in agent-process mode was dead until recovery, and recovery could not succeed either.

Root cause

The Phase 4.5 rewrite replaced the five provider services with LlmChatRunner, and LlmServiceProxy was updated to resolve it:

// TelegramSearchBot.LLMAgent/Service/LlmServiceProxy.cs
_chatRunner = serviceProvider.GetRequiredService<LlmChatRunner>();

The worker builds its own DI container manually (LLMAgentProgram.BuildServices) and was only updated for the renamed provider-API classes (OpenAiModelApi …), not for the newly introduced LlmChatRunner / LlmProviderRegistry. Constructing the executor therefore throws

InvalidOperationException: No service for type 'TelegramSearchBot.Service.AI.LLM.LlmChatRunner' has been registered.

and because AgentLoopService.ProcessTaskAsync resolves the executor outside its try, the exception escapes RunAsync and kills the worker process.

CI stayed green because the only worker-DI test asserted that ResponsesModelApi still resolves.

Fix

• Register LlmProviderRegistry and LlmChatRunner (singletons, matching the main-process DI scan lifetimes) in LLMAgentProgram.BuildServices.
• Add regression test BuildServices_RegistersAgentTaskExecutor, which resolves IAgentTaskExecutor (constructs LlmServiceProxy → LlmChatRunner). Fails before the fix, passes after.

Validation

• dotnet build TelegramSearchBot.sln -c Release → 0 errors
• dotnet test TelegramSearchBot.LLM.Test → 267/267 passed
• dotnet test TelegramSearchBot.Test → 445 passed, 1 skipped (pre-existing skip)

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 16999324-f1e4-4835-b336-121efc400839

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

PR Check Report

Summary

• PR: fix(llm): register LlmChatRunner in LLMAgent worker DI #391
• Branch: fix/agent-worker-llmchatrunner-di <- master
• Event: pull_request
• Commit: 0b4635d

Test Results

Platform	Status	Details
Ubuntu	Passed	Tests passed, artifacts uploaded
Windows	Passed	Tests passed, artifacts uploaded

Code Quality

• Code formatting check
• Security vulnerability scan
• Dependency analysis
• Code coverage collection

Test Artifacts

• Test results artifacts count: 2
• Code coverage uploaded to Codecov

Links

• PR Page
• Actions Logs

---

This report is auto-generated by GitHub Actions