Agent tries to run code on bash script instructions

[michnovka]

What happened?

This has no sense. I do not know how agent can decide whether to run the code or not, maybe the prompt should be adjusted so that librechat knows what code to run and what not. If I turn off run code option it works like this:

Steps to Reproduce

1. enable run code in agent
2. ask a question which gives you bash code in reply as instructions
3. see it fail to try and execute code

What browsers are you seeing the problem on?

No response

Relevant log output

Screenshots

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[danny-avila]

bash is not one of the supported runtimes/languages.

The agent already receives plenty of instructions from the tool definition but you can better steer the agent with additional instructions for this.

It's also not clear to see what exactly it tried to do. Please share any relevant logs from the ./logs directory as well as a screenshot for the expanded view of "Analyzing"

[michnovka]

I am not sure how to get any meaningful logs. I have DEBUG_LOGGING=true in .env but I see only this:

2025-02-02T15:54:16.117Z debug: [BaseClient] Loading history:
{
  conversationId: "607405b4-1f34-43f1-bce0-9c99fa4568aa",
  parentMessageId: "00000000-0000-0000-0000-000000000000",
}
2025-02-02T15:54:16.119Z debug: [BaseClient] Context Count (1/2)
{
  remainingContextTokens: 199979,
  maxContextTokens: 200000,
}
2025-02-02T15:54:16.119Z debug: [BaseClient] Context Count (2/2)
{
  remainingContextTokens: 199979,
  maxContextTokens: 200000,
}
2025-02-02T15:54:16.119Z debug: [BaseClient] tokenCountMap:
{
  dd636b2a-9873-4c43-baec-ecb41da8a5e5: 18,
}
2025-02-02T15:54:16.120Z debug: [BaseClient]
{
  promptTokens: 21,
  remainingContextTokens: 199979,
  payloadSize: 1,
  maxContextTokens: 200000,
}
2025-02-02T15:54:16.120Z debug: [BaseClient] tokenCountMap
{
  dd636b2a-9873-4c43-baec-ecb41da8a5e5: 18,
}
2025-02-02T15:54:16.120Z debug: [BaseClient] userMessage
{
  messageId: "dd636b2a-9873-4c43-baec-ecb41da8a5e5",
  parentMessageId: "00000000-0000-0000-0000-000000000000",
  conversationId: "607405b4-1f34-43f1-bce0-9c99fa4568aa",
  sender: "User",
  text: "how to create 4k RSA key on centos 8 ?",
  isCreatedByUser: true,
  tokenCount: 18,
}
2025-02-02T15:54:16.121Z debug: [api/server/controllers/agents/client.js] chatCompletion
{
  baseURL: undefined,
  // 1 payload(s)
  payload: [{"role":"user","content":"how to create 4k RSA key on centos 8 ?"}],
}
2025-02-02T15:54:16.135Z debug: [saveConvo] api/app/clients/BaseClient.js - saveMessageToDatabase #saveConvo
2025-02-02T15:54:24.542Z debug: [spendTokens] conversationId: 607405b4-1f34-43f1-bce0-9c99fa4568aa | Context: message | Token usage:
{
  promptTokens: 3953,
  completionTokens: 116,
}
2025-02-02T15:54:24.544Z debug: [spendTokens] conversationId: 607405b4-1f34-43f1-bce0-9c99fa4568aa | Context: message | Token usage:
{
  promptTokens: 4096,
  completionTokens: 206,
}
2025-02-02T15:54:24.555Z debug: [spendTokens] No transactions incurred against balance
2025-02-02T15:54:24.560Z debug: [spendTokens] No transactions incurred against balance
2025-02-02T15:54:24.563Z debug: [saveConvo] api/app/clients/BaseClient.js - saveMessageToDatabase #saveConvo
2025-02-02T15:54:24.573Z debug: [AgentController] Request closed
2025-02-02T15:54:29.627Z debug: [spendTokens] conversationId: 607405b4-1f34-43f1-bce0-9c99fa4568aa | Context: title | Token usage:
{
  promptTokens: 461,
  completionTokens: 33,
}
2025-02-02T15:54:29.627Z debug: [spendTokens] conversationId: 607405b4-1f34-43f1-bce0-9c99fa4568aa | Context: title | Token usage:
{
  promptTokens: 714,
  completionTokens: 39,
}
2025-02-02T15:54:29.628Z debug: [saveConvo] api/server/services/Endpoints/agents/title.js
2025-02-02T15:54:29.631Z debug: [spendTokens] No transactions incurred against balance
2025-02-02T15:54:29.632Z debug: [spendTokens] No transactions incurred against balance

The Analyzing log was expanded already, see highlighted what is the collapsible section